FSTL4 interaction with Gαi proteins activates the Akt-mTOR pathway to drive malignant phenotypes in non-small cell lung cancer Authors: Jiaxin Li, Wanghong Qi, Shansong Gao, Yifei Zhu, Jianhua Zha Published: 2026 ABSTRACT AbstractNon-small cell lung cancer (NSCLC) presents a critical challenge clinically; thus, identifying novel drivers and reliable prognostic biomarkers is imperative. Here, we investigated the clinical relevance, cell-type-specific localization, and mechanistic role of Follistatin-like 4 (FSTL4) in NSCLC. Analysis of The Cancer Genome Atlas (TCGA) and local resected tissue samples show that FSTL4 expression is significantly elevated in NSCLC tissues compared to normal lung tissues, and serves as an independent prognostic indicator for reduced overall survival. Single-cell RNA sequencing confirmed that FSTL4 expression is predominantly and consistently localized to the malignant epithelial cell population across primary tumors and metastatic niches in lung adenocarcinoma (LUAD). Functionally, stable knockdown or CRISPR/Cas9-mediated knockout of FSTL4 across multiple NSCLC cell types suppressed key malignant phenotypes, including cell proliferation, and migration, and inducing G1-phase cell cycle arrest and triggering intrinsic apoptosis. This cytotoxic dependency was selective to malignant cells, as normal lung epithelial cells remained unaffected. Conversely, ectopic overexpression of FSTL4 enhanced these aggressive characteristics. Mechanistically, FSTL4 exerts its pro-oncogenic effects through regulating the Akt-mTOR signaling cascade. FSTL4 physically interacts with the inhibitory G-protein subunits Gɑi1 and Gɑi3, and the co-depletion of these subunits inhibited FSTL4-mediated Akt-mTOR activation. Crucially, the functional effects of FSTL4 silencing in primary human NSCLC cells were rescued by the introduction of a constitutively active Akt1 (caAkt1, S473D) construct, confirming Akt as the essential required downstream transducer. In vivo validation further demonstrated that FSTL4 silencing profoundly attenuated NSCLC xenograft tumor growth by reducing proliferation and accelerating apoptosis, mirroring the diminished Akt-mTOR activation observed in vitro. Collectively, these comprehensive findings unequivocally establish that elevated FSTL4 expression promotes malignant phenotypes of NSCLC possibly by activating Gɑi1/3-Akt-mTOR cascade.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12964-026-03001-z. Introduction Non-small cell lung cancer (NSCLC) is the most prevalent form of lung cancer [1, 2], constituting approximately 80–85% of all diagnoses and representing a leading cause of cancer-related mortality worldwide [3–5]. This disease is heterogeneous, primarily comprising adenocarcinoma, squamous cell carcinoma, and large cell carcinoma, each defined by distinct molecular characteristics [3, 6–8]. Despite a multimodal standard of care that includes surgery, platinum-based chemotherapy, radiation, and immunotherapy, the five-year survival rate for patients with advanced or metastatic disease remains suboptimal [3, 6, 7]. A principal challenge impeding therapeutic success is the high incidence of both intrinsic and acquired drug resistance [6, 7, 9, 10], which underscores a critical need to elucidate the molecular drivers of NSCLC to develop more effective treatment strategies [6, 7, 9–11]. The emergence of targeted therapy has significantly altered the treatment paradigm for select subsets of NSCLC patients [3, 7, 10, 12]. This research aligns with the primary objectives of our group's work [13–16]. Follistatin-like 4 (FSTL4) is a glycoprotein within the follistatin family, a group of proteins distinguished by a cysteine-rich follistatin domain [17]. While its definitive biological functions are yet to be fully elucidated, emerging evidence indicates that FSTL4 could be possibly involved in fundamental cellular processes. FSTL4 knockout (KO) mice showed heightened hippocampal excitability and long-term potentiation (LTP), enhanced extinction of aversive memory that depends on brain-derived neurotrophic factor (BDNF) signaling, with anti-BDNF blocking these effects [18]. Consistent with the function of other follistatin family members, FSTL4 is presumed to regulate key signaling pathways [17]. Suzuki et al., showed that FSTL4 negatively regulated memory extinction by interacting with BDNF and inhibiting its role in synaptic plasticity [18]. In mouse brain, FSTL4 is upregulated by corticosterone-induced endoplasmic reticulum stress, contributing to reduced neuronal viability and synaptic-related protein levels [19]. Methylation of the FSTL4 gene, among other genes, was found to have a significant correlation with multiple clinical indicators of dyslipidemia in Mexican populations [20]. By applying a novel gene-based Hardy–Weinberg equilibrium test (gene-HWT) to genotype data from Japanese patients, researchers identified FSTL4 as a significant multi-cancer germline factor, showing particularly strong associations with colorectal and prostate cancers [21]. Aflatoxin B1 induces malignant transformation in liver cells by disrupting a complex regulatory network, specifically identifying FSTL4 as one of the key genes whose expression is co-regulated by DNA methylation and miRNAs to drive hepatocellular carcinoma [22]. Researchers also identified genetic variants at the FSTL4 locus that contribute to an inverse risk relationship between Alzheimer’s disease and human cancers (breast and prostate), potentially mediated by shared microRNA regulatory networks [23]. Interestingly, a recent study shows that neuronal FSTL4 serves as a critical negative regulator of glioma growth by inhibiting the maturation of BDNF [24]. Despite these diverse insights, the expression and function of FSTL4 have not yet been studied in NSCLC. This study, therefore, is designed to systematically investigate the expression, biological functions, and underlying molecular mechanisms of FSTL4 in NSCLC. The central objective is to evaluate its potential as a novel prognostic biomarker and a future therapeutic target in this malignancy. Materials and methods Reagents All reagents for cell culture, including serum, medium and antibiotics, were sourced from Hyclone (Logan, UT). The following fluorescent dyes were purchased from Thermo Fisher Scientific (Carlsbad, CA, USA): JC-1, DAPI, EdU, TUNEL. The BrdU ELISA Kit was acquired from Roche Diagnostics (Indianapolis, IN, USA). Transwell chambers were obtained from Corning Incorporated (New York, NY). CCK-8, trypan blue staining kit and other reagents were obtained from Sigma-Aldrich (Shanghai, China). Cells We obtained the human lung adenocarcinoma (LUAD) cell line, A549, from the Shanghai Institute for Biological Sciences (Shanghai, China) and cultured it in RPMI-1640 medium with 10% FBS [13–16]. Primary human non-small cell lung cancer (NSCLC) cells (pNSCLC-1, pNSCLC-2, pNSCLC-3, derived three patients) and primary lung epithelial cells (pEpi1 and pEpi2, from two doners) were isolated from consenting patients and donors, as previously detailed [14–16]. These cells were maintained under conditions described in earlier studies [13–16]. We routinely tested the cell lines for mycoplasma and microbial contamination, and performed genetic stability profiling (short tandem repeat profiling) to ensure their authenticity. All protocols involving human cells were approved by the Ethics Committee of Nanchang University and complied with the Declaration of Helsinki. Human tissues We collected twenty (n = 20) pairs of human NSCLC tumor ("T") and adjacent normal lung epithelial ("N") tissues from patients with advanced-stage (III-IV) lung adenocarcinoma (LUAD) at our institution (reported in our early studies [13, 14]). Upon surgical removal, the tissue specimens were immediately frozen in liquid nitrogen to preserve their integrity. They were later homogenized in a lysis buffer containing protease inhibitors. All patient materials and clinical data were handled in accordance with strict ethical guidelines. The lung cancer samples were collected from patients who provided written-informed consent, and the study was approved by the Ethics Committee of the First Affiliated Hospital of Nanchang University. The research adhered to the principles outlined in the Declaration of Helsinki. Single-cell RNA sequencing As previously detailed [25–27], single-cell RNA sequencing (scRNA-seq) data were processed using Seurat v5.1.0 in the R computing environment. The dataset, GSE131907, was obtained from Kim et al. through the GEO database [28]. We implemented strict quality control measures to remove low-quality cells and technical artifacts. Data normalization was performed using the SCTransform method to correct for technical variations. Repaired principal component analysis (rPCA) was used for batch effect integration. For dimensionality reduction and visualization of cellular heterogeneity, we employed uniform manifold approximation and projection (UMAP). Cell annotations were directly sourced from the original publications to ensure consistency with established classifications. Gene silencing To silence FSTL4, NSCLC cells or lung epithelial cells were grown to 60% confluence in complete medium with polybrene. They were then transduced with FSTL4 shRNA constructs (Genechem, Shanghai, China) using lentivirus at a multiplicity of infection (MOI) of 11. After a 48-h incubation, we used puromycin selection to create stable shRNA-expressing cell lines within six passages. Two different shRNA sequences were used to target FSTL4: shFSTL4-seq1 (targeting: CTCCTTCACTGAAAGCAATCAATAC) and shFSTL4-seq2 (targeting: CTCCTCCGATTTGACGATTACAACA). Control cells were transduced with a scrambled shRNA (shC) lentivirus [13]. We continuously monitored FSTL4 mRNA and protein levels in all cells. The Gαi1 and Gαi3 shRNA lentiviruses were provided by Dr. Cao [29–32]. The silencing procedure for both was identical to the method described above. CRISPR/Cas9-mediated gene editing CRISPR/Cas9 method was employed to generate FSTL4 knockout (KO) NSCLC cells. NSCLC cells were first cultured to 60% confluence in complete medium supplemented with polybrene. They were then transduced with a lentiviral vector expressing the Cas9 enzyme, as previously described [13, 14]. After establishing stable Cas9-expressing cells, we infected them with a lentiviral vector containing a CRISPR/Cas9-FSTL4-KO puromycin resistance cassette. This construct carried one of two single guide RNAs (sgRNAs) targeting distinct FSTL4 loci: koFSTL4-sg1 (targeting TGTTGTAATCGTCAAATCGG, PAM: AGG) or koFSTL4-sg2 (targeting ACTGACTATGAAGCGCCCGT, PAM: CGG), both from Genechem (Shanghai, China). Stable FSTL4 KO cell clones were isolated using puromycin selection in 96-well plates. The successful knockout of FSTL4 was confirmed by targeted deep sequencing of the sgRNA site and Western blotting to verify the absence of the protein. Control cells were generated using the same procedure with a lentiviral vector encoding a CRISPR/Cas9-construct with non-sense sgRNA ("sgC"), as previously reported [13]. The CRISPR/Cas9 Gαi1 and Gαi3 knockout constructs were also provided by Dr. Cao [29–32], with the procedures identical to the method described above. Gene overexpression To achieve robust and stable overexpression of FSTL4, we transduced NSCLC cells with a lentiviral construct (GV369, Genechem) [13, 14] that contained the full-length FSTL4 cDNA sequence (hFSTL4[NM_015082.2]). The multiplicity of infection (MOI) was 11. Following the infection, cells were cultured for 48 h to allow for efficient viral integration and gene expression. Stable cells with constitutive FSTL4 overexpression were then established through a stringent puromycin selection process over six passages. To ensure consistent and faithful FSTL4 overexpression, we periodically monitored its mRNA and protein levels in these stable cells. Gαi1 and Gαi3 overexpression using the lentiviral constructs provided by Dr. Cao [29–32] was achieved using the same protocols. Generation of a constitutively-active Akt1 mutant NSCLC cells were transduced with a lentivirus carrying a constitutively-active Akt1 mutant (caAkt1, S473D), which was reported early [13] and provided by Dr. Cao [33]. After a 48-h incubation, puromycin was added for an additional 96 h to select for stably transduced cells. The presence of caAkt1 was confirmed through Western blotting. Examining cellular functions and gene/protein expression To assess cellular functions and gene/protein expression, we cultured equal quantities of genetically modified NSCLC cells and normal lung epithelial cells under optimized conditions for specific durations. A comprehensive panel of assays was employed to characterize various cellular functions, including immunofluorescence staining, cell viability (CCK-8 assay), proliferation (nuclear EdU/DAPI staining, BrdU ELISA, colony formation assay), migration (Transwell assay), cell cycle progression [Propidium Iodide (PI)-FACS], apoptosis (Caspase-3/−9 activity assays, cytosol cytochrome C ELISA, nuclear TUNEL/Hoechst 33,342 staining), and cell death (trypan blue exclusion). The detailed methodologies for these assays have been previously published [13–16]. We also performed gene expression analysis using quantitative PCR (qRT-PCR) and protein expression profiling via Western blotting, as described in our prior studies [13]. Nuclear proteins were isolated using a commercial nuclear extraction kit (Thermo-Fisher Scientific) according to the manufacturer’s instructions, involving the sequential lysis of the cytoplasmic membrane followed by high-salt extraction of the intact nuclei. The detailed procedures for immunoprecipitation (IP) assay were also described in earlier studies, testing 600 μg total protein lysates per sample [34]. Uncropped Western blotting images are provided in Figure S1. Xenograft studies Detailed protocols for these studies were previously published [13, 35]. We used immunodeficient nude mice (5–6 weeks old, 18.1–19.1 g) obtained from SLAC Laboratory Animal Co., Ltd. (Shanghai, China) and housed them in a specific pathogen-free (SPF) environment at the Animal Facility of Nanchang University. We used an equal number of male and female mice for the study. To establish subcutaneous xenograft tumors, we injected six million pNSCLC-1 cells into the right flanks of each mouse. Tumor growth was monitored three weeks after initial cell injection (Day-0). Tumor dimensions, recorded every six days, were measured to calculate tumor volume using standard formulas [13, 35]. All animal experiments were approved by the Animal Ethics Committee of Nanchang University and conducted in accordance with their guidelines. Immunohistochemistry (IHC) and tissue fluorescence staining Tumor specimens were fixed in formalin, embedded in paraffin, and sectioned into 4 µm thick slices. As reported previously [13, 35], we followed standard procedures for deparaffinization and rehydration, followed by antigen retrieval in citrate buffer. Non-specific binding was blocked before staining. The slides were stained for nuclear Ki-67, p-Akt or FSTL4 according to protocols reported early [13]. Statistical analysis To reduce experimental bias, investigators were blinded to group assignments throughout the in vitro studies. All experiments were performed in five independent replicates, which yielded consistent results. We used one-way analysis of variance (ANOVA) followed by Tukey's multiple comparison test for comparisons involving multiple groups. For two-group comparisons, we used the Student's t-test. Data are presented as the mean ± standard deviation (SD). P-values of < 0.05 were considered statistically significant. Results Elevated FSTL4 expression in NSCLC is associated with poor clinical outcomes First, as determined by an analysis of The Cancer Genome Atlas (TCGA) data [13, 14], FSTL4 expression is significantly higher in NSCLC tumor tissues compared to normal lung tissues. As shown in Fig. 1A, a comparison of FSTL4 expression levels revealed a marked increase in tumor samples compared to the normal lung tissues. A paired analysis further confirmed this finding, showing that FSTL4 expression was consistently and significantly upregulated in tumor tissues compared to matched normal tissues from the same patients, as shown in Fig. 1B. Importantly, Elevated FSTL4 expression serves as a strong prognostic indicator for reduced patient survival. Kaplan–Meier survival analysis in Fig. 1C demonstrates that patients with high FSTL4 expression have a significantly shorter overall survival (OS) compared to those with low expression. The hazard ratio (HR) for OS was 1.45 (95% CI: 1.08–1.95), indicating a heightened risk of mortality (P = 0.014). Patients with high FSTL4 expression exhibited a highly comparable trend toward reduced Disease-Specific Survival (DSS), maintaining a similar hazard ratio magnitude to that of overall survival (HR = 1.37, 95% CI: 0.95–1.98) (Fig. 1D). While this trend did not reach the strict threshold for statistical significance (P = 0.093) (Fig. 1D)-likely due to reduced statistical power from fewer recorded disease-specific events and the inherent complexities of precise mortality attribution in large clinical cohorts-the parallel trajectories of the OS and DSS curves suggest a strong clinical relevance. The prognostic significance of FSTL4 was further validated in stratified analyses. In male NSCLC patients, high FSTL4 expression was a significant predictor of poor OS, with a hazard ratio of 1.94 (95% CI: 1.28–2.93, P = 0.002; Fig. 1E). Moreover, high FSTL4 expression was consistently associated with worse OS regardless of age. Both patients aged 65 or younger (HR = 1.63, 95% CI: 1.07–2.50, P = 0.024; Fig. 1F) and those older than 65 (HR = 1.64, 95% CI: 1.10–2.45, P = 0.015; Fig. 1G) with high FSTL4 expression experienced significantly shorter overall survival. These findings underscore the role of FSTL4 as a potential biomarker for poor prognosis in NSCLC. Elevated FSTL4 expression in NSCLC is associated with poor clinical outcomes. The expression of FSTL4 in non-small cell lung cancer (NSCLC) was analyzed using public data from The Cancer Genome Atlas (TCGA). FSTL4 expression levels in NSCLC tumor tissues were compared to those in normal lung tissues (A), and in a subset of NSCLC patients with paired normal and tumor samples (B). The relationship between FSTL4 expression and patient survival was assessed using Kaplan–Meier analysis, evaluating both overall survival (OS) (C) and disease-specific survival (DSS) (D). Further stratified analyses were performed to examine the prognostic value of FSTL4 in different subgroups based on gender (male) (E) and age (≤ 65 and > 65) (F and G). Hazard ratios (HR) and 95% confidence intervals (CI) were calculated for all survival analyses. TPM stands for transcripts per million. *Indicates P < 0.05 (A and B) High FSTL4 expression correlates with the activation of important signaling pathways To identify genes whose expression is linked to FSTL4 in NSCLC, a comprehensive correlation analysis was performed using data from TCGA. As shown in Fig. 2A, genes with a Pearson correlation coefficient (cor-r) greater than 0.3 and an adjusted P-value less than 0.05 were selected as being significantly co-expressed with FSTL4 in NSCLC tissues. This stringent filtering process isolated a specific set of genes that are positively correlated with the expression of FSTL4 (Fig. 2A). Further functional and pathway enrichment analysis of these co-expressed genes provided insight into their biological roles. Figure 2B displays the top Gene Ontology (GO) Biological Process (BP) terms that were significantly enriched. The analysis revealed a strong enrichment for pathways related to cAMP-mediated signaling, G protein-coupled receptor (GPCR) signaling, and glucose metabolic processes. These findings suggest a direct functional link between FSTL4 and these critical cellular processes. This biological context was further supported by analysis using the Reactome pathway database. As depicted in Fig. 2C, the enriched pathways included Insulin Processing, Peptide Hormone Metabolism, and G alpha Signaling Events. These pathways are all directly connected to the signaling and metabolic functions identified in the GO analysis. High FSTL4 expression correlates with the activation of key signaling and metabolic pathways. A comprehensive co-expression analysis was conducted using TCGA NSCLC gene expression data to identify genes correlated with FSTL4 (A). A stringent threshold of a Pearson correlation coefficient (cor-r) > 0.3 and an adjusted P-value < 0.05 was applied to isolate a set of significantly co-expressed genes (A). The selected gene list was then subjected to functional and pathway enrichment analysis. Gene Ontology (GO) Biological Process (BP) terms were analyzed to identify enriched biological functions (B). Additionally, pathways were analyzed using the Reactome database to provide further context on the molecular networks associated with FSTL4 co-expression (C) Single-cell analysis reveals FSTL4 expression is predominantly localized to malignant epithelial cells across primary and metastatic sites in LUAD Lung adenocarcinoma (LUAD) is the predominant histological subtype of non-small cell lung cancer (NSCLC), accounting for approximately 40% of all NSCLC cases [6, 10]. It is also the most common type of lung cancer found in non-smokers and is frequently associated with specific genetic mutations, such as EGFR and KRAS [6, 10]. To investigate the cell-type-specific expression of FSTL4 in NSCLC, we analyzed single-cell RNA sequencing data from a published dataset (GSE131907), with cell annotations provided by the original authors [28]. The analysis consistently showed that FSTL4 expression is primarily localized within the malignant epithelial cell population across various tumor microenvironments in LUAD. As depicted in Fig. 3A and B, FSTL4 was most highly expressed in the malignant epithelial cell cluster of advanced primary LUAD tumors. This suggests that FSTL4 expression is closely associated with the malignant epithelial cells themselves. The same pattern was observed in single-cell data from LUAD brain metastases. The results in Fig. 3C and D show that FSTL4 is predominantly and highly expressed within the epithelial cell cluster in these metastatic LUAD samples. Analysis of data from LUAD-affected lymph nodes also confirmed this finding. As shown in Fig. 3E and F, FSTL4 expression is highly concentrated in the malignant epithelial cell cluster. Notably, its expression is even more pronounced in metastatic lymph nodes compared to non-metastatic ones, suggesting a potential role in the metastatic process (Fig. 3E and F). The single-cell analysis of LUAD pleural fluids further validated this cell-type-specific expression. Figure 3G and H demonstrate that FSTL4 is highly expressed in the malignant epithelial cell population found in these pleural fluid samples. These results consistently indicate that FSTL4 expression is predominantly found within the malignant epithelial cell population across primary and metastatic sites in LUAD, highlighting its potential role in the biology of the tumor cells themselves. Single-cell analysis reveals FSTL4 expression is predominantly localized to malignant epithelial cells across primary and metastatic sites in LUAD. Single-cell RNA sequencing data from the dataset GSE131907 was used to investigate the cell-type-specific expression of FSTL4 in lung adenocarcinoma (LUAD). FSTL4 expression levels were mapped onto UMAP plots to visualize its distribution across different cell populations within primary LUAD tumors (A and B), brain metastases (“mBrain”, C and D), affected lymph nodes (E and F), and pleural fluid (PE) samples (G and H) FSTL4 expression is markedly upregulated in locally-resected NSCLC tissues and various NSCLC cell types The expression status of FSTL4 was next evaluated in locally-resected NSCLC specimens (T, as reported in our early studies [13, 14]) relative to matched, adjacent normal lung tissues (N). qRT-PCR analysis, performed on twenty paired tissue samples (N and T, n = 20), revealed that FSTL4 mRNA levels were markedly and significantly increased in the NSCLC tumor cohort when compared to the normal tissues (Fig. 4A). This transcriptional upregulation was subsequently confirmed at the protein level. Western blot analysis demonstrated that the expression of the FSTL4 protein was concomitantly elevated in the tumor samples from four representative patients compared to the normal epithelial controls, with GAPDH serving as the essential internal loading control (Fig. 4B). The densitometric quantification of all twenty paired samples demonstrated that the expression of the FSTL4 protein was concomitantly and significantly elevated in the tumor tissues compared to the normal epithelial controls (Fig. 4C). To corroborate these clinical findings and define the expression pattern in an in vitro system, we assessed FSTL4 expression across several NSCLC and normal lung cell types. FSTL4 mRNA levels were observed to be significantly higher in the three primary NSCLC cell type (pNSCLC-1, − 2, and − 3, as reported in our early studies [13, 14]) and the widely utilized immortalized A549 cell line when compared to the normal human lung epithelial cell types (pEpi1 and pEpi2, from two doners, also reported in our early studies [13, 14]) (Fig. 4D). Furthermore, Western blot combined with densitometric quantification verified a parallel and statistically significant increase in FSTL4 protein content across all tested NSCLC cell types relative to lung epithelial cell types (Fig. 4E). These data establish a consistent, robust overexpression of FSTL4 at both the messenger RNA and protein levels in NSCLC clinical specimens and cultured cell models. FSTL4 expression is markedly upregulated in locally-resected NSCLC tissues and various NSCLC cell types. qRT − PCR analysis of FSTL4 mRNA expression in twenty paired NSCLC tumor tissues (T) relative to adjacent normal lung tissues (N) (A). Representative Western blot demonstrating FSTL4 protein expression in four pairs of NSCLC tumor and normal tissues; GAPDH served as the loading control (B). Densitometric quantification of FSTL4 protein levels across all twenty paired NSCLC tumor tissues relative to normal tissues (C). qRT − PCR analysis of FSTL4 mRNA expression in three primary NSCLC cell types (pNSCLC-1, − 2, and − 3) and the immortalized A549 cell line compared to normal human lung epithelial cell types (pEpi1 and pEpi2) (D). Western blot analysis and corresponding densitometric quantification verifying FSTL4 protein expression across the cell types analyzed in (E); GAPDH served as the loading control. Data are expressed as mean ± standard deviation (SD). Statistical significance was tested relative to “N” tissues or “pEpi1” cells (*P < 0.05) FSTL4 silencing suppresses malignant phenotypes across various NSCLC cells To elucidate the functional implications of FSTL4 overexpression, a gene silencing strategy was implemented by utilizing lentiviral short hairpin RNA (shRNA) constructs targeting FSTL4 in primary pNSCLC-1 cells. Transfection with two independent shRNA sequences (shFSTL4-seq1 and shFSTL4-seq2, with non-overlapping sequences) resulted in a robust and substantial reduction in both endogenous FSTL4 mRNA and protein abundance when compared against parental control (Pare) and non-targeting shRNA control (shC) groups (Fig. 5A and B). Crucially, the specificity of the gene silencing was confirmed, as the expression levels of the control gene FSTL1 were not perturbed at either the transcriptional or protein level (Fig. 5A and B). The silencing of FSTL4 dramatically compromised the intrinsic proliferative capacity of primary pNSCLC-1 cells. As shown, FSTL4 shRNA significantly impaired colony formation efficiency over a 14-day incubation period (Fig. 5C) and caused a substantial decrease in cell viability after 96 h, as determined by the CCK − 8 assay (Fig. 5D). Furthermore, cell proliferation was compromised, evidenced by substantial reductions in both nuclear BrdU incorporation and EdU staining across shFSTL4-treated pNSCLC-1 cells (Fig. 5E and F). In addition to proliferative effects, FSTL4 depletion substantially suppressed the in vitro metastatic capability of NSCLC cells. Cell migration (assessed via Transwell assay without Matrigel coating) was markedly attenuated in primary pNSCLC-1 cells following FSTL4 knockdown, with measurements taken at the 24-h time point (Fig. 5G). Moreover, the PI-flow cytometric cell cycle analysis demonstrated that FSTL4 silencing elicited a pronounced accumulation of cells in the G1 checkpoint phase (Fig. 5H). This arrest was correlated with a significant reduction in the population of cells progressing into the S (synthesis) phase (Fig. 5H), indicating that FSTL4 is important for G1/S transition of pNSCLC-1 cells. FSTL4 silencing suppresses malignant phenotypes across various NSCLC cells. Lentiviral shRNA was used to silence FSTL4 in pNSCLC-1 cells using two independent sequences (shFSTL4 − seq1 and shFSTL4 − seq2). qRT − PCR and Western blot analysis confirmed the specific reduction of FSTL4 mRNA and protein levels without affecting FSTL1 expression, compared to parental (Pare) and non-targeting shRNA (shC) controls (A and B). The impact of FSTL4 silencing on cellular functions was assessed via: colony formation assay over 14 days (C); CCK − 8 assay to measure cell viability (D); Nuclear BrdU and EdU incorporation assays to determine proliferative capacity (E and F). Transwell assays were used to evaluate in vitro cell migration (without Matrigel) (G). Flow cytometric analysis assessed cell cycle distribution, focusing on G1/S phases (H). FSTL4 mRNA knockdown specificity (using shFSTL4 − seq2) was confirmed in pNSCLC-2, pNSCLC-3, and A549 cells using qRT − PCR assays, with no change in FSTL1 mRNA expression (I and J). Subsequent analyses measured cell viability by CCK − 8 assay (K), nuclear EdU incorporation to assess proliferation (L), and Transwell assays for cell migration capability (M) across these three additional cell types. All quantitative data are presented as the mean ± standard deviation (SD), with a sample size of n = 5. Statistical significance was determined by comparison to the "shC" control group, with a P-value less than 0.05 considered significant (*P < 0.05). Non-significant differences are indicated as "n.s." (P > 0.05). Each experiment was independently replicated five times, with consistent results observed across all biological repeats. Scale bars in the figure represent 100 µm To establish the broad relevance of these findings, the functional consequence of FSTL4 depletion was independently validated across a panel of additional NSCLC cell models, specifically two distinct primary cell types (pNSCLC-2 and pNSCLC-3) and the A549 immortalized cell line. Parallel knockdown experiments in pNSCLC-2, pNSCLC-3, and A549 cells showed that the application of shFSTL4-seq2 resulted in robust FSTL4 mRNA silencing, without affecting FSTL1 mRNA expression (Fig. 5I and J). FSTL4 silencing consistently led to profound and uniform decreases in cellular viability across all three NSCLC cell models (Fig. 5K). This reduction in viability was substantiated by nuclear EdU staining assays, which revealed a universal and significant suppression of DNA synthesis (Fig. 5L). Furthermore, the in vitro migratory capabilities of all three NSCLC cells were distinctly and uniformly attenuated upon the reduction of FSTL4 expression (Fig. 5M). Taken in concert, the results from this comprehensive array of in vitro functional assays conclusively establish FSTL4 as an important driver of the aggressive malignant phenotypes inherent to NSCLC cells. FSTL4 silencing triggers intrinsic apoptotic pathways in NSCLC cells Given the suppression of cell proliferation and viability as well as arrest of cell cycle observed upon FSTL4 silencing, we proceeded to investigate whether these consequences were attributable to the induction of apoptosis in NSCLC cells. Utilizing the pNSCLC-1 cells, we observed that the employment of two distinct shRNA constructs targeting FSTL4 (shFSTL4-seq1 and shFSTL4-seq2, see Fig. 5) resulted in activation of key executioner caspases. Specifically, the enzymatic activities of both Caspase − 3 and Caspase − 9 were substantially amplified following FSTL4 knockdown at 72 h (Fig. 6A and B). ELISA assay results showed that FSTL4 silencing induced significant increase in the cytosolic accumulation of Cytochrome C (Fig. 6C). Furthermore, analysis using the potential-sensitive JC-1 fluorescent probe revealed that FSTL4 knockdown caused a rapid and significant disruption of mitochondrial membrane potential (ΔΨm​), evidenced by a distinct shift from red J-aggregates to green J-monomers (Fig. 6D). The nuclear TUNEL staining visually revealed an increase in the percentage of cells exhibiting apoptotic nuclei (Fig. 6E). Notably, FSTL4 silencing induced a modest increase in apoptosis in pNSCLC-1 cells, with the apoptotic rate remaining below 20% after 72 h. The trypan blue staining assay further demonstrated that FSTL4 knockdown caused moderate (< 20%) yet significant pNSCLC-1 cell death (Fig. 6F). To decisively confirm that the observed cytotoxicity was functionally dependent upon apoptosis, NSCLC − 1 cells were treated with the broad-spectrum pan-caspase inhibitor, zVAD-fmk, or the caspase-3 specific inhibitor zDEVD-fmk. Pre-treatment with these inhibitors effectively inhibited the reduced viability and cell death phenotypes induced by FSTL4 knockdown by shFSTL4-seq2 (Fig. 6G and H). FSTL4 silencing triggers intrinsic apoptotic pathways in NSCLC cells. pNSCLC-1 cells were transfected with two distinct shRNA constructs (shFSTL4 − seq1 and shFSTL4 − seq2). Enzymatic activity assays were performed to measure the activation of Caspase-3 and Caspase-9, compared to parental (Pare) and non-targeting shRNA (shC) controls (A and B). The accumulation of Cytochrome C in the cytosol was quantified using an ELISA kit (C). Mitochondrial membrane potential (ΔΨm) was assessed via JC-1 fluorescent probe analysis (D). Nuclear TUNEL staining was used to quantify the percentage of apoptotic cells (E). Cell death was quantified using the Trypan blue staining assay (F). pNSCLC-1 cells with shFSTL4 − seq2 or shC were treated with the pan-caspase inhibitor zVAD − fmk (Casi, 50 μM) or the Caspase-3 specific inhibitor zDEVD − fmk (Cas3i, 50 μM) before assessing viability and cell death (G and H). The induction of apoptosis was validated across pNSCLC-2, pNSCLC-3, and A549 cell types using shFSTL4 − seq2 or shC by measuring mitochondrial depolarization via JC-1 assay (I), TUNEL staining to quantify apoptotic nuclei (J), and total cell death (L), with representative immunofluorescence images of pNSCLC-2 cells co-stained for FSTL4 (green) and cleaved Caspase-9 (red), with DAPI (blue) highlighting the nuclei shown (K). Specificity was demonstrated by confirming FSTL4 and FSTL1 expression levels in normal human lung epithelial cells (pEpi1 and pEpi2) following shFSTL4 − seq2 or shC transfection (M and N). CCK − 8 assay and Trypan blue assays were used to assess the viability and cell death in normal epithelial cells following FSTL4 knockdown (O and P). All quantitative data are presented as the mean ± standard deviation (SD), with a sample size of n = 5. Statistical significance was determined by comparison to the "shC" control group, with a P-value less than 0.05 considered significant (*P < 0.05). Non-significant differences are indicated as "n.s." (P > 0.05). # P < 0.05 vs. “Veh” (0.2% DMSO) group (G and H). Each experiment was independently replicated five times, with consistent results observed across all biological repeats. Scale bars in the figure represent 100 µm The apoptotic analysis was extended to a validation panel comprising two additional primary NSCLC cell types (pNSCLC-2 and pNSCLC-3) and the A549 immortalized cell line. FSTL4 knockdown by shFSTL4-seq2 (see Fig. 5) consistently induced mitochondrial depolarization (Fig. 6I) and resulted in a moderate increase in apoptotic nuclei (Fig. 6J) across all three NSCLC cell models. We performed dual-labeling immunofluorescence (IF) staining in pNSCLC-2 cells. In control cells (shC), robust expression of FSTL4 (green) was associated with negligible levels of cleaved-Caspase-9 (red), indicating low baseline apoptosis. In contrast, stable knockdown of FSTL4 (shFSTL4-seq2) resulted in increased cleaved-Caspase-9 immunoreactivity in pNSCLC-2 cells (Fig. 6K). Moreover, overall cell death was elevated in all FSTL4-silenced NSCLC cell types (Fig. 6L). In stark contrast to the effects observed in NSCLC cells, the functional consequences of FSTL4 silencing were assessed in two distinct normal human lung epithelial cell isolates (pEpi1 and pEpi2, see Fig. 4). While FSTL4 (but not FSTL1) expression was successfully reduced in these normal epithelial cells using the same shFSTL4-seq2 procedure (Fig. 6M and N), this reduction did not translate into a compromised phenotype. Specifically, the viability, as assessed by the CCK − 8 assay, remained unaffected in both pEpi1 and pEpi2 cells following FSTL4 knockdown (Fig. 6O). Concurrently, there was no increase in cell death in either normal epithelial cell type (Fig. 6P). Thus, NSCLC cells exhibit a unique dependence on FSTL4. Genetic knockout of FSTL4 via CRISPR/Cas9 technology suppresses malignant phenotypes in primary NSCLC cells To further confirm the necessity of FSTL4 in NSCLC malignancy, we utilized the CRISPR/Cas9 system to generate stable FSTL4 knockout (koFSTL4) in pNSCLC-1 cells. Western blot analysis confirmed a near-complete loss of FSTL4 protein expression in the koFSTL4 pNSCLC-1 cell groups using two independent single-guide RNA constructs (sg1 and sg2), with no concurrent effect on the expression of the control FSTL1 (Fig. 7A). Genetic ablation of FSTL4 systematically inhibited all tested malignant phenotypes in pNSCLC-1 cells. Compared to the scrambled guide control (sgC), koFSTL4 pNSCLC-1 cells demonstrated a significant reduction in DNA synthesis, as measured by EdU incorporation (Fig. 7B). Furthermore, FSTL4 knockout resulted in a substantial attenuation of in vitro cell migration (Fig. 7C). The loss of FSTL4 was sufficient to induce moderate programmed cell death, which was detected by an increase in the percentage of apoptotic nuclei via TUNEL staining (Fig. 7D). To confirm these effects, the pNSCLC-2 cells were assessed. The CRISPR/Cas9 system using the sg1 construct resulted in near complete FSTL4 protein loss, without affecting FSTL1 protein expression (Fig. 7E). As shown, koFSTL4 also resulted in a significant inhibition in cell proliferation (EdU incorporation) (Fig. 7F). The metastatic potential was similarly suppressed, with substantial inhibition of in vitro cell migration (Fig. 7G). Apoptotic analysis further confirmed the phenotype, showing a moderate increase in apoptotic nuclei (TUNEL staining) (Fig. 7H). Genetic knockout of FSTL4 via CRISPR/Cas9 technology suppresses malignant phenotypes in primary NSCLC cells. CRISPR/Cas9 gene editing was used to generate stable FSTL4 knockout (koFSTL4) in pNSCLC-1 cells using two independent guide RNAs (koFSTL4-sg1 and koFSTL4-sg2). Western blot confirmed the near-complete loss of FSTL4 protein without affecting FSTL1 expression compared to the control construct with non-sense sgRNA (sgC) (A). The functional consequences of FSTL4 knockout in pNSCLC-1 cells were assessed via nuclear EdU incorporation to measure cell proliferation (B). Transwell assays determined cell migration (C). Apoptosis was assessed by measuring the percentage of apoptotic nuclei via TUNEL staining (D). The effects were validated in pNSCLC-2 cells, starting with Western blot confirmation of FSTL4 protein loss using koFSTL4-sg1 and lack of effect on FSTL1 protein expression (E). Functional assays were then performed, including nuclear EdU incorporation for proliferation (F), Transwell assays for cell migration (G), and TUNEL staining for apoptotic nuclei (H). Data are presented as mean ± SD. * indicates a significant difference vs. sgC cells. “n.s.” stands for non-statistical difference. Each experiment was independently replicated five times, with consistent results observed across all biological repeats. Scale bars in the figure represent 100 µm Ectopic overexpression of FSTL4 enhances malignant phenotypes in primary NSCLC cells To functionally validate that the elevated expression of FSTL4 actively contributes to the malignant characteristics of NSCLC, we established stable cell models exhibiting FSTL4 overexpression (oeFSTL4) in pNSCLC-1 cells. These models were generated using a single lentiviral construct, resulting in two independent stable cell selections (oeFSTL4-sl1 and oeFSTL4-sl2). We first confirmed the successful genetic manipulation in pNSCLC-1 cells, observing a significant increase in both FSTL4 mRNA and protein levels via qRT-PCR and Western blot, respectively (Fig. 8A and B). Importantly, this forced increase in FSTL4 had no discernible effect on the basal expression of the control gene, FSTL1 (Fig. 8A and B). The functional analysis of pNSCLC-1 cells demonstrated that forced FSTL4 overexpression is a powerful driver of cell proliferation. Specifically, we observed a marked increase in cell viability (assessed by CCK − 8 assay) (Fig. 8C) and enhancement of DNA synthesis, evidenced by both nuclear BrdU and EdU incorporation assays (Fig. 8D and E). Furthermore, the Transwell assays showed that oeFSTL4 significantly amplified the in vitro cell migration (Fig. 8F). Ectopic overexpression of FSTL4 enhances malignant phenotypes in primary NSCLC cells. pNSCLC-1 cells were stably transfected with a single lentiviral construct to achieve FSTL4 overexpression (oeFSTL4), yielding two independent stable cell selections (oeFSTL4-sl1 and oeFSTL4-sl2). qRT − PCR and Western blot analysis confirmed the increased FSTL4 mRNA and protein levels, respectively, with no resulting change in FSTL1 expression, compared to the vector control cells (Vec) (A and B). The functional consequences of FSTL4 overexpression in pNSCLC-1 cells were measured using the CCK − 8 assay for cell viability (C). BrdU and EdU incorporation assays assessed DNA synthesis and proliferative capacity (D and E). Transwell assays were utilized to determine in vitro cell migration (F). FSTL4 mRNA overexpression was confirmed in pNSCLC-2, pNSCLC-3, and A549 cells, using the same lentiviral construct (G), with confirmation that FSTL1 mRNA was unaffected (G). Functional analyses across these three validation cell models included CCK − 8 assays for cell viability (H), EdU incorporation assays for DNA synthesis (I), and Transwell assays for in vitro cell migration (J). All quantitative data are presented as the mean ± standard deviation (SD), with a sample size of n = 5. Statistical significance was determined by comparison to the "Vec" control group, with a P-value less than 0.05 considered significant (*P < 0.05). Non-significant differences are indicated as "n.s." (P > 0.05). Each experiment was independently replicated five times, with consistent results observed across all biological repeats. Scale bars in the figure represent 100 µm To ensure the generality of these oncogenic observations, the findings were validated across two additional primary cell types (pNSCLC-2 and pNSCLC-3) and the A549 cell line. In all these cell models, the application of the same lentiviral overexpression construct (oeFSTL4) resulted in robust FSTL4 mRNA overexpression (Fig. 8G), without affecting FSTL1 mRNA expression (Fig. 8G). FSTL4 overexpression consistently induced elevated cell viability (Fig. 8H), heightened DNA synthesis (EdU incorporation) (Fig. 8I), and enhanced in vitro migration (Fig. 8J) when compared to the vector control (Vec). These comprehensive experiments provide evidence that the elevated expression of FSTL4 functions as a sufficient molecular factor for driving the malignant phenotypes of NSCLC cells. Akt-mTOR axis is a key signaling pathway mediating the oncogenic functions of FSTL4 in NSCLC The Akt-mTOR cascade is critically implicated in NSCLC due to its ubiquitous role in promoting cellular proliferation, survival, and angiogenesis, thus driving tumor progression and resistance to conventional therapies [36–39]. Given that TCGA bioinformatic studies have revealed the potential involvement of FSTL4 in key signaling pathways in NSCLC, we further explored its role in modulating Akt-mTOR signaling. Western blot analysis unequivocally demonstrated that genetic depletion of FSTL4, either through stable knockdown (by shFSTL4-seq2, see Figs. 5 and 6) or CRISPR knockout (via koFSTL4-sg2, see Fig. 7) in pNSCLC-1 cells, resulted in a substantial decrease in the phosphorylation levels of Akt (specifically at S473) and its critical downstream effector, S6K1 (at T389) (Fig. 9A). Conversely, stable overexpression of FSTL4 (oeFSTL4, see Fig. 8) produced the reciprocal effect, leading to a marked elevation in the phosphorylation of both Akt and S6K1 (Fig. 9B). These findings establish FSTL4 as an important regulator of the Akt − mTOR signaling cascade in NSCLC. Akt-mTOR axis is a key signaling pathway mediating the oncogenic functions of FSTL4 in NSCLC. pNSCLC-1 cells with genetic depletion of FSTL4 (stable knockdown, using shFSTL4 − seq2; or CRISPR-induced knockout, via koFSTL4 − sg2) were analyzed for the listed proteins, compared to the corresponding controls, shC or sgC (A and C). Conversely, pNSCLC-1 cells with stable FSTL4 overexpression (oeFSTL4) were analyzed for the phosphorylation of Akt and S6K1 (B). pNSCLC-1 cells silenced for FSTL4 (shFSTL4 − seq2) were transfected with a constitutively active Akt1 plasmid (caAkt1, S473D) or a vector control. Western blot was used to confirm the restoration of Akt and S6K1 phosphorylation (D). The functional consequences of Akt restoration were assessed by measuring cell proliferation via nuclear EdU incorporation (E), migratory capacity via Transwell assay (F), and apoptosis via TUNEL staining and quantification (G). All quantitative data are presented as the mean ± standard deviation (SD), with a sample size of n = 5. Statistical significance was determined by comparison to the "shC"/“sgC” control group, with a P-value less than 0.05 considered significant (*P < 0.05) (A-C). Non-significant differences are indicated as "n.s." (P > 0.05). # P < 0.05 (D-G). Each experiment was independently replicated five times, with consistent results observed across all biological repeats. Scale bars in the figure represent 100 µm To determine the effects of FSTL4 on other well-established oncogenic cascades in NSCLC, we examined the activation status of the MAPK/ERK, STAT3, and Wnt-β-catenin pathways. Western blot analysis confirmed a significant reduction in FSTL4 protein levels in both the shFSTL4 and koFSTL4 pNSCLC-1 cells compared to their respective controls (shC and sgC). However, the loss of FSTL4 did not lead to detectable changes in the phosphorylation of Erk1/2 (Thr202/Tyr204) or STAT3 (Tyr705), suggesting that the MAPK and STAT3 signaling cascades are not primary downstream targets of FSTL4 in this context (Fig. 9C). Furthermore, we assessed the canonical Wnt/β-catenin signaling pathway by measuring the nuclear β-catenin levels. Quantitative analysis of the nuclear fraction showed no significant difference in nuclear β-catenin in levels between FSTL4-depleted pNSCLC-1 cells and control cells (Fig. 9C). Collectively, these data demonstrate that FSTL4 deficiency does not globally suppress oncogenic signaling but rather exhibits specificity (Akt-mTOR) in NSCLC cells. To confirm that the activation of Akt is the requisite event mediating FSTL4's functional role, we conducted mechanistic rescue experiments. pNSCLC-1 cells, in which FSTL4 expression was silenced (via shFSTL4 − seq2), were transiently rescued by transfecting a constitutively active Akt1 plasmid (S473D, caAkt1). Western blot confirmed that caAkt1 expression successfully restored the phosphorylation status of both Akt and S6K1 (Fig. 9D), without affecting FSTL4 expression (Fig. 9D). Functionally, the restoration of Akt activity significantly abrogated the anti-NSCLC cell effects induced by FSTL4 suppression. Specifically, co-expression of caAkt1, when compared to shFSTL4-treated cells alone, led to a substantial reversal of the reduction in cell proliferation (measured by EdU incorporation) (Fig. 9E), re-established the migratory capacity (Fig. 9F), and significantly suppressed the induction of cell apoptosis (assessed by TUNEL staining) (Fig. 9G). Thus, Akt-mTOR axis is a key signaling pathway mediating the oncogenic functions of FSTL4 in NSCLC. FSTL4 associates with Gαi1 and Gαi3 to mediate downstream Akt − mTOR pathway activation in NSCLC cells Gαi proteins, integral members of the heterotrimeric G protein family, transduce extracellular signals by primarily inhibiting adenylyl cyclase, thereby attenuating intracellular cAMP production, and can additionally regulate various other downstream effectors [40–42]. Indeed, the inhibitory G-protein subunits Gαi1 and Gαi3 are established upstream mediators of Akt − mTOR activation across diverse cellular contexts [29–32, 34, 43–53]. The observed co-expression of FSTL4 with genes involved in cAMP-mediated signaling and Gα protein signaling pathways from TCGA bioinformatic analyses (see Fig. 2) prompted our investigation into the mechanistic relationship between FSTL4 and Gαi proteins. Immunoprecipitation (IP) assays conducted in pNSCLC-1 cells, pNSCLC-2 cells and N1 human NSCLC tissues (see Fig. 4) confirmed the association: FSTL4 co-precipitated with Gαi3 when using an anti-Gαi3 antibody, which also co-precipitated with Gαi1 (Fig. 10A), and, both Gαi1 and Gαi3 were successfully pulled down using an anti-FSTL4 antibody (Fig. 10A). These findings provide evidence for the physical complex formation between FSTL4 and both Gαi1 and Gαi3 within the NSCLC tissues and cells. FSTL4 associates with Gαi1 and Gαi3 to mediate downstream Akt − mTOR pathway activation in NSCLC cells. Immunoprecipitation (IP) assays were performed in pNSCLC-1 cells, pNSCLC-2 cells, and fresh N1 human NSCLC tissues to examine the physical association between FSTL4 and the inhibitory G-protein subunits Gαi1 and Gαi3; IP was conducted using antibodies against FSTL4 or Gαi3 (A). Western blot confirmation of dual Gαi1 and Gαi3 depletion via shRNA (shGαi1/3) or CRISPR knockout (koGαi1/3) in pNSCLC-1 cells stably overexpressing FSTL4 (oeFSTL4 − sl1) (B). Western blot analysis showing the phosphorylation levels of Akt (S473) and S6K1 (T389) in oeFSTL4 − sl1 cells following dual Gαi1/Gαi3 silencing or KO (C). Western blot analysis of p − Akt (S473) and p − S6K1 (T389) levels after ectopic co-overexpression of Gαi1 and Gαi3 in oeFSTL4 − sl1 pNSCLC-1 cells (D and E). All quantitative data are presented as the mean ± standard deviation (SD), with a sample size of n = 5. "shC + koC" stands for control cells with scramble control shRNA plus non-sense sgRNA CRISPR/Cas9 construct. Statistical significance was determined by comparison to the "shC + koC"/“Vec” control group, with a P-value less than 0.05 considered significant (*P < 0.05). Non-significant differences are indicated as "n.s." (P > 0.05). Each experiment was independently replicated five times, with consistent results observed across all biological repeats To ascertain the functional requirement of these Gαi1/3 proteins, we tested their role in FSTL4-mediated Akt − mTOR activation. We utilized pNSCLC-1 cells stably overexpressing FSTL4 (oeFSTL4 − sl1) and subsequently performed shRNA silencing of both Gαi1 and Gαi3 (shGαi1/3) or CRISPR knockout (ko) of both Gαi1 and Gαi3 (koGαi1/3) (Fig. 10B). Western blot analysis demonstrated that the robust elevation in Akt (S473) and S6K1 (T389) phosphorylation induced by FSTL4 overexpression was largely inhibited following the dual knockdown or knockout of Gαi1 and Gαi3 (Fig. 10C). Conversely, ectopic overexpression of Gαi1 and Gαi3 (jointly) in control pNSCLC-1 cells effectively enhanced phosphorylation status of Akt and S6K1 (Fig. 10D and E) in oeFSTL4 − sl1 pNSCLC-1 cells. Thus, FSTL4 engages in a functional protein complex with Gαi1 and Gαi3 to mediate the activation of the Akt − mTOR signaling cascade, which ultimately governs NSCLC malignancy. FSTL4 silencing inhibits subcutaneous NSCLC cell xenograft growth in nude mice To validate the pro-oncogenic role of FSTL4 in a physiological context, we established an in vivo subcutaneous xenograft model. Each mouse was grafted with six million pNSCLC-1 cells, stably expressing either a control short hairpin RNA (shC) or FSTL4 knockdown (shFSTL4 − seq2), through subcutaneous injection to the right flank. Within three weeks, the recordings were started (labelled as Day-0). As shown, the trajectory of tumor growth was profoundly attenuated in mice injected with shFSTL4 − seq2 pNSCLC-1 cells (Fig. 11A). The estimated tumor growth rate per day was also substantially diminished in the FSTL4-silenced pNSCLC-1 xenografts (Fig. 11B). FLST4 silencing also resulted in markedly lower tumor weights compared to the shC control group at Day-42 (Fig. 11C). Importantly, this inhibition of pNSCLC-1 xenograft progression was not attributed to systemic toxicity, as the body weights of the recipient mice remained statistically indifferent between the two groups throughout the experimental period (Fig. 11D). FSTL4 silencing inhibits subcutaneous NSCLC cell xenograft growth in nude mice. pNSCLC-1 cells stably expressing either control short hairpin RNA (shC) or FSTL4 knockdown (shFSTL4 − seq2) were subcutaneously injected into nude mice. Within three weeks, the recordings were started (labelled as Day-0). The trajectory of tumor growth every six days over 42 days (Day-0 to Day-42, A) and the estimated tumor growth rate per day (B) were recorded. Final tumor weights were measured at Day − 42 (C). Body weights of recipient mice were monitored throughout the experiment (D). At Day-18 and Day-36, one xenograft per group were isolated, the molecular analysis of excised xenograft tissues confirmed successful FSTL4 depletion at mRNA and protein levels by Western blot and qRT-PCR, with FSTL1 expression used as a control (E and F). IHC staining further verified FSTL4 protein silencing (G). Western blot analysis showed that FSTL4 silencing significantly reduced the phosphorylation of Akt (S473) and S6K1 (T389) (H and I). Functional analysis revealed reduced cell proliferation, as indicated by diminished Ki67 IHC signal (J), and increased apoptosis, evidenced by elevated Caspase − 3 activity (K) and corresponding PARP1 cleavage (L). Quantitative data are expressed as the mean ± standard deviation (SD). Statistical significance was determined by comparison to the shC control group; a P-value less than 0.05 was considered significant (*P < 0.05), while non-significant differences were denoted as "n.s." (P > 0.05). For panels A–D, each experimental group included a total of ten mice (n = 10). The data presented in panels E–L were derived from five randomly selected tissue samples per xenograft (n = 5). All images include a scale bar of 100 µm Molecular analysis of the excised xenograft tissues was performed to confirm the successful genetic manipulation and investigate the underlying mechanism. Western blotting and qRT-PCR confirmed that FSTL4 expression was successfully depleted at both the mRNA and protein levels in the shFSTL4 − seq2 group, while FSTL1 expression remained unaffected (Fig. 11E and F). Immunohistochemical (IHC) results in Fig. 11G further verified silencing of FSTL4 protein in shFSTL4 − seq2 pNSCLC-1 xenograft slides. This confirmed depletion of FSTL4 corresponded to a dramatic reduction in the phosphorylation of Akt (S473) and its downstream target, S6K1 (T389) (Fig. 11H and I), consistent with the molecular mechanism established in vitro. In addition, shFSTL4 − seq2 pNSCLC-1 xenografts showed a pronounced reduction in cell proliferation, as indicated by a diminished signal for the proliferation marker Ki67 (Fig. 11J). Concurrently, a significant increase in apoptosis was observed, evidenced by elevated Caspase-3 activity (Fig. 11K) and the corresponding cleavage of PARP1 (Fig. 11L). Collectively, these comprehensive in vivo results conclusively demonstrate that FSTL4 is an essential factor for NSCLC xenograft growth, acting by maintaining the active Akt − mTOR signaling axis, and promoting cellular proliferation and inhibits apoptosis. Discussion While targeted therapies against oncogenic drivers such as EGFR and ALK have advanced the treatment of NSCLC, their efficacy is constrained by resistance and is limited to specific patient subsets [6, 10, 54–56]. A significant cohort of NSCLC patients lacks such actionable mutations, restricting their therapeutic avenues [6, 10, 54–56]. This clinical reality creates a pressing imperative to identify and validate novel molecular targets for the development of new strategies that can benefit a wider patient population and overcome existing therapeutic hurdles [6, 10, 54–56]. Our research identifies FSTL4 as a novel and clinically significant therapeutic target in NSCLC. Analysis of the TCGA database, further validated by qRT-PCR and Western blot of locally-treated clinical specimens, consistently revealed significant FSTL4 upregulation in NSCLC tumor tissues relative to normal lung tissue. This elevated expression is not merely correlational; it holds substantial prognostic value. Kaplan–Meier survival analysis demonstrated a robust association between high FSTL4 levels and diminished overall survival, establishing FSTL4 as an independent predictor of poor clinical outcomes. Moreover, single-cell RNA sequencing analysis localized FSTL4 expression predominantly to malignant epithelial cells in primary tumors, metastases, and pleural effusions, confirming that its pro-tumorigenic activity originates from the cancer cells themselves. We rigorously established the functional dependence of NSCLC malignancy on FSTL4 through a series of genetic manipulation studies. The use of both shRNA-mediated silencing and CRISPR/Cas9-mediated knockout in various primary and established NSCLC cell types led to a consistent suppression of key malignant phenotypes, including reduced proliferation, migration, alongside the induction of G1/S cell cycle arrest and apoptosis. Conversely, ectopic overexpression of FSTL4 exacerbated these aggressive oncogenic traits. The clinical relevance of these observations was confirmed in vivo, where a subcutaneous xenograft model showed that FSTL4 silencing profoundly attenuated tumor growth. Critically, FSTL4 depletion in normal lung epithelial cells did not affect viability or trigger cell death, indicating a cancer-specific dependency and suggesting a promising therapeutic window for FSTL4-targeted interventions. To elucidate the molecular mechanisms driving these phenotypes, we identified the Akt-mTOR signaling cascade as a critical downstream effector of FSTL4. This pathway is a known contributor to cell growth, survival, and proliferation in NSCLC [36, 39, 57–59]. Our results revealed a direct correlation between FSTL4 expression levels and the phosphorylation status of the key pathway components Akt and S6K1; genetic depletion of FSTL4 attenuated Akt-mTOR signaling, while its overexpression enhanced it. We confirmed the functional significance of this axis through rescue experiments, in which the introduction of a constitutively active Akt1 (caAkt1) largely inhibited the anti-tumorigenic effects of FSTL4 silencing. These results collectively demonstrate that FSTL4 executes its oncogenic functions principally via activation of the Akt-mTOR pathway. The Gαi subfamily of heterotrimeric G proteins encompasses three principal isoforms: Gαi1, Gαi2, and Gαi3. Elevated levels of these Gαi proteins are a recurring feature in several human cancers, where they exert a profound influence on tumorigenesis and disease advancement [33, 34, 49, 60, 46]. Contemporary research highlights the indispensable functions of Gαi1 and Gαi3 in transducing signals from a broad spectrum of receptor tyrosine kinases (RTKs), such as EGFR, FGFR, KGFR, c-Kit, TrkB, and VEGFR2 [30, 32, 34, 50, 51, 61]. Notably, these Gαi isoforms are essential intermediaries for the activation of the pro-oncogenic Akt-mTOR pathway initiated by these cell surface receptors [30, 32, 34, 50, 61, 51]. Extending beyond their well-characterized RTK interactions, Gαi1 and Gαi3 also form associations with diverse non-RTK receptors. Prominent instances include their engagement with LGR4 in response to R-spondin 3 (RSPO3), CD146 upon Netrin-1 binding, the IL-4-stimualted interleukin-4 (IL-4) receptor, and Toll-like receptor 4 (TLR4) when stimulated by lipopolysaccharide (LPS) [29, 31, 52, 62]. These results establish Gαi proteins, Gαi1 and Gαi3, as pivotal regulators of the Akt-mTOR signaling cascade, thereby reinforcing their status as key oncogenic drivers and presenting promising avenues for targeted cancer therapies [33, 34, 49, 46, 60]. Here, the insight mechanisms of FSTL4-induced Akt-mTOR cascade activation in NSCLC cells were explored. Prompted by bioinformatic analyses implicating Gɑ protein and cAMP signaling of FSTL4-coexpressing genes, we explored a potential interaction with inhibitory Gα (Gαi) proteins, which are known mediators of Akt-mTOR activation (see above). Through immunoprecipitation assays, we discovered a novel physical association between FSTL4 and the Gαi subunits Gαi1 and Gαi3 in both NSCLC cells and patient tissues. The functional significance of this interaction was established when the dual silencing or knockout of Gαi1/3 in FSTL4-overexpressing cells largely inhibited the FSTL4-driven phosphorylation of Akt and S6K1. These results support a model wherein FSTL4 forms a functional complex with Gαi1/3 to mediate the downstream activation of the Akt-mTOR signaling cascade, thereby driving NSCLC malignancy. Therefore, FSTL4 is a newly identified oncogenic driver in NSCLC that is overexpressed in tumors and linked to poor prognosis. It promotes cancer progression by activating the Gαi1/3-dependent Akt-mTOR pathway, making it a potential target for new NSCLC therapies. Supplementary Information Supplementary Material 1. Authors’ contributions All authors designed this study, carried out all in vitro cellular and in vivo animal experiments, and performed bioinformatical All the listed authors contributed to drafting the article, critically revising it for significant intellectual content, and gave final approval for the version submitted to the journal. Funding This research was funded by the National Natural Science Foundation for Youths (No. 82302962), Jiangxi Natural Science Foundation (20242BAB25509), Jiangxi Natural Science Foundation for Youths (20232BAB216066), the Research Start-up Fund of the Yangfan Project from the First Affiliated Hospital of Nanchang University (RSC-0061), the Science and Technology Research Project of Jiangxi Provincial Department of Education (No. GJJ2400814), the Science and Technology Plan Project of Jiangxi Administration of Traditional Chinese Medicine (No. 2023B1224) and the Start Up Fund for Doctoral Research of Jiangxi University of Traditional Chinese Medicine(No. 2022BSZY013). Data availability All data are with the figures and its supplementary files. Declarations Ethics approval and consent to participate All methods were carried out in accordance with Declaration of Helsinki and other relevant guidelines and regulations. Lung cancer samples and primary cells were collected from written-informed consent patients with approval from the Ethics Committee of the First Affiliated Hospital of Nanchang University (NO. (2024) CDYFYYLK (08–011)). All animal experiment procedures were performed in accordance with the Animal Ethics Committee of Nanchang University (CDYFY-IACUC-202406QR026). Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Jiaxin Li, Wanghong Qi and Shansong Gao contributed equally to this work. Contributor Information Yifei Zhu, Email: zhuyifei1119_rjh@163.com. Jianhua Zha, Email: zhajianhua@ncu.edu.cn. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement All data are with the figures and its supplementary files.