Introduction

LUAD represents the predominant histological subtype of non-small cell lung cancer with unfavorable clinical outcomes. Its incidence and mortality continue to rise worldwide, making it a leading cause of cancer-related death [1, 2]. Although surgical resection, targeted therapies and immune checkpoint inhibitors have substantially improved survival in selected patient subsets, the overall therapeutic benefit remains limited. This constraint largely stems from the remarkable metabolic plasticity of tumor cells, the establishment of immune evasion programs and the progressive emergence of treatment resistance [3, 4]. Increasing evidence suggests that these processes are not independent but are tightly interconnected; however, the molecular basis underlying their coordination remains insufficiently understood.

Tumor metabolic reprogramming is a hallmark of cancer, with enhanced glycolysis-commonly referred to as the “Warburg effect”-representing a central mechanism by which cancer cells sustain rapid proliferation and survival under hypoxic and nutrient-limited conditions [5]. Extensive evidence indicates that glycolysis not only supplies energy and biosynthetic precursors to tumor cells, but also profoundly remodels the TME through lactate accumulation, extracellular acidification, and the signaling functions of metabolic intermediates. This remodeling suppresses effector T-cell activity while promoting the infiltration of immunosuppressive populations, such as M2 macrophages [6, 7]. Clinically, high expression of glycolysis-related genes in LUAD patients correlates with reduced overall survival and poor therapeutic response, highlighting enhanced glycolysis as both a driver of tumor progression and a determinant of adverse prognosis [8]. Despite these advances, the regulatory molecules that coordinate glycolytic reprogramming with immune modulation in LUAD remain poorly defined, limiting the development of effective metabolism-targeted therapeutic strategies.

Recent studies have highlighted metabolic intervention as a promising approach to overcome immunotherapy resistance and enhance antitumor efficacy [9]. Fluvastatin (Fluv), a classical HMG-CoA reductase inhibitor, exerts lipid-lowering effects and, beyond this role, has been shown in multiple tumor models to interfere with cholesterol and isoprenoid metabolic pathways, thereby suppressing tumor growth, disrupting metabolic homeostasis, and modulating the immune microenvironment [10, 11]. In parallel, accumulating evidence indicates that certain photosensitizers used in Photodynamic therapy (PDT), while inducing oxidative stress, may also interfere with tumor metabolic pathways [12,13,14,15], including the regulation of glycolytic activity. These findings suggest that integrating metabolic inhibition with PDT may provide a potential strategy to simultaneously target tumor metabolism and immune evasion. However, the molecular mechanisms underlying such metabolic–immune coupling, as well as the mediators involved, remain largely unexplored. Notably, tumor-derived exosomes (TDEs) have emerged as critical regulators of intercellular communication within the TME, capable of transferring chemokines and signaling molecules that influence immune cell recruitment, differentiation, and functional status, thereby contributing to immunosuppressive remodeling [16,17,18].

In this study, we systematically investigated the interplay between glycolytic reprogramming and immune regulation in LUAD. We established a glycolysis-related prognostic model to evaluate its association with patient outcomes and TME characteristics, and identified DEPDC1 as a potential regulator linking metabolic and immune processes. Furthermore, we developed a cRGD-functionalized nanophotosensitizer (cRGD-TPA-DCR-NP, termed NPL) and combined it with Fluv to establish a metabolism-targeted combinatorial strategy (Fluv + NPL). This approach suppresses glycolysis and exosomal CCL2 levels, potentially through DEPDC1-related mechanisms, and is associated with improvements in the tumor immune microenvironment (Scheme 1). These findings provide new insights into the metabolic–immune regulatory network in LUAD and highlight a promising therapeutic strategy targeting this axis.

Scheme 1
Scheme 1
Full size image

The schematic illustrates the therapeutic strategy combining Fluv with cRGD-TPA-DCR-based photodynamic nanoparticles for the treatment of LUAD. The cRGD-TPA-DCR-NP, prepared via nanoprecipitation and composed of cRGD, DSPE-PEG2000-Mal, DSPE-PEG2000, and TPA-DCR. The combination of fluvastatin and light-irradiated cRGD-TPA-DCR nanoparticles (Fluv + NPL) induced immunogenic cell death (ICD), as evidenced by increased calreticulin (CRT) surface exposure and enhanced HMGB1 release, which may contribute to increased effector T-cell infiltration. Mechanistically, Fluv + NPL suppressed DEPDC1-mediated NF-κB signaling, thereby not only inhibiting PI3K–AKT–mTOR pathway activation and reducing glycolytic activity in LUAD cells, but also decreasing exosomal CCL2 levels. This leads to reduced expression of the glycolysis-related proteins (GLUT1, HK2, LDHA), decreased lactate accumulation in the TME, and facilitating effector T-cells infiltration. Compared with Controls, Fluv + NPL also suppresses CCL2 levels in LUAD-derived exosomes, thereby decreasing the proportion of M2 macrophages while increasing M1 macrophages. Collectively, these effects improve the TME (M1↑, M2↓, effector T cells↑) and synergistically inhibit LUAD growth. In particular, the identified glycolytic and exosome-mediated immune regulatory axes associated with DEPDC1 enhance our understanding of the mechanisms linking glycolytic metabolism to immune regulation and may facilitate the development of DEPDC1-targeted therapies

Methods and materials

Detailed methods are listed in Supplemental Information.

Results

1. Glycolysis-based risk stratification predicts prognosis in LUAD

To evaluate the prognostic significance of glycolysis in LUAD, we performed a systematic analysis using the KEGG_GLYCOLYSIS_GLUCONEOGENESIS gene set (M11521) derived from the Molecular Signatures Database (MSigDB). Univariate Cox regression analysis was first conducted to identify glycolysis-related genes significantly associated with overall survival. The expression landscape of these prognostic glycolytic genes was visualized using a heatmap, revealing that the majority were upregulated in LUAD tissues (Fig. 1A). Based on the subset of glycolysis-associated genes significantly correlated with patient survival, we subsequently constructed a glycolysis-related prognostic risk model for LUAD (Fig. S1A-B). The predictive performance of this model was evaluated in both the training and testing cohorts, demonstrating robust stability and reliable prognostic discrimination (Fig. S1C-D). To determine whether the glycolysis risk score independently predicts clinical outcome, we performed multivariate Cox regression analyses incorporating the risk score and clinicopathological variables in both cohorts. The results confirmed that the glycolysis-derived risk score serves as an independent prognostic factor for LUAD patients (Fig. 1C-D).

We next compared the expression patterns of model-constituting genes between the high- and low-risk groups and examined their associations with clinicopathological characteristics (Fig. 1B). Patients were stratified into high- and low-risk subgroups according to the median glycolysis risk score, followed by Kaplan-Meier survival analysis. Across the entire cohort, as well as in the training and testing sets, patients in the high-risk group exhibited significantly reduced overall survival (Fig. S1E, Fig. 1F–G), indicating robust prognostic discrimination. Given the emerging association between metabolic reprogramming and genomic instability, we next assessed tumor mutation burden (TMB) and found that tumors in the high-risk group exhibited significantly elevated TMB levels (Fig. 1E). To characterize the tumor microenvironmental landscape, ESTIMATE analysis was performed. Compared with the low-risk group, high-risk tumors displayed markedly reduced ImmuneScore, StromalScore and ESTIMATEScore (Fig. 1H-J), suggesting decreased immune and stromal cell infiltration and an overall more immunologically suppressed TME.

To further evaluate the robustness and generalizability of the glycolysis-related prognostic risk model constructed based on the TCGA-LUAD cohort, we validated it using two independent external GEO datasets (GSE30219 and GSE37745). The results showed that in both independent cohorts, patients in the high-risk group exhibited significantly worse overall survival than those in the low-risk group (GSE30219, p < 0.001; GSE37745, p = 0.003) (Fig. S1F and Fig. S1I). Further multivariate Cox regression analysis demonstrated that, after adjusting for clinical variables including age and sex, the risk score remained an independent prognostic factor (GSE30219, HR = 1.214, p = 0.005; GSE37745, HR = 1.239, p = 0.020) (Fig. S1G and Fig. S1J). In both external cohorts, the ESTIMATEScore were higher in the low-risk group compared with the high-risk group, and the differences were statistically significant (Fig. S1H and Fig. S1K). In summary, through survival analysis, multivariate Cox regression, and ESTIMATEscore evaluation in two independent external datasets, we systematically validated the prognostic model. Collectively, the glycolysis-based risk model effectively predicts adverse clinical outcomes in LUAD and is closely associated with elevated TMB and an immune-suppressed TME characterized by reduced immune and stromal infiltration. However, these findings primarily reflect clinical correlations, and the glycolytic regulators underlying this high-risk phenotype, as well as their therapeutic tractability, remain to be elucidated. To investigate this further, LUAD cells were treated with the metabolic modulator Fluv, and integrated transcriptomic and targeted metabolomic analyses were performed to systematically characterize Fluv-induced changes in glycolysis-associated gene expression and metabolism.

Fig. 1
Fig. 1
Full size image

Construction and validation of a glycolysis-related prognostic model in LUAD. (A) Differential expression of glycolysis-related genes between tumor and normal tissues based on TCGA-LUAD data. (B) Heatmap of model gene expression stratified by glycolysis risk score, showing the distribution of clinical features (stage, age, sex, risk score) across groups. (C-D) Forest plots from multivariate Cox regression assessing the prognostic value of clinical variables and glycolysis risk score in the test set (C) and training set (D). (E) Differences in TMB between high- and low-risk groups. (F-G) Validation of prognostic performance of the glycolysis risk score in the test set (F) and training set (G), including risk score distribution, survival status, Kaplan-Meier curves, and heatmap of model gene expression. (H-J) Violin plots of the TME scores stratified by glycolysis risk groups calculated using the ESTIMATE algorithm: ImmuneScore (H), ESTIMATEScore (I), and StromalScore (J)

2. Fluvastatin treatment identifies DEPDC1-associated glycolytic features and prognostic relevance

To identify metabolically responsive candidate genes underlying glycolysis-associated risk in LUAD, we performed transcriptomic profiling of A549 cells treated with Fluv or Control. Differential expression analysis revealed that Fluv treatment resulted in 1,206 downregulated and 844 upregulated genes (Fig. S1L). Pathway enrichment analysis of differentially expressed genes indicated significant enrichment in metabolic pathways and the PI3K-AKT signalling cascade (Fig. 2A). To further refine candidate genes relevant to LUAD pathogenesis, we identified differentially expressed genes between tumor and adjacent normal tissues using three independent algorithms (DESeq2, edgeR and limma), yielding 2,932 intersecting genes (Fig. S1M). We then intersected Fluv vs. Control differentially expressed genes (DEGs), LUAD-associated differentially expressed genes and an immune-related gene set. This integrative analysis identified 55 overlapping candidates, including CCL2, XDH, PCK2, PTGER4, APLN, BIRC5, E2F1, DUSP1 and JUND (Fig. 2B). qPCR validation in A549 and H1299 cells confirmed that Fluv treatment suppressed the mRNA expression of CCL2, PTGER4, APLN, BIRC5 and E2F1, while upregulating XDH, PCK2, DUSP1 and JUND (Fig. 2D), consistent with transcriptomic profiling. To specifically interrogate glycolysis-associated candidate genes, we further intersected Fluv vs. Control DEGs, LUAD differentially expressed genes and the glycolysis gene set. Five candidate genes were identified: DEPDC1, CDK1, AURKA, HMMR and KIF20A (Fig. 2C). qPCR validation demonstrated that Fluv treatment markedly reduced the mRNA expression of all five genes in LUAD cells (Fig. 2E), in agreement with RNA-seq data. To further prioritize candidate glycolytic genes, we performed survival analyses using TCGA-LUAD data to evaluate the association between gene expression and overall survival (OS). High expression levels of DEPDC1 (Fig. S1N), CDK1 (Fig. S1O), AURKA (Fig. S1P), HMMR (Fig. S1Q) and KIF20A (Fig. S1R) were all associated with poorer survival outcomes. Notably, among these five genes, DEPDC1 exhibited the highest hazard ratio (HR = 1.91), indicating the strongest association with adverse prognosis in LUAD patients. Based on its pronounced prognostic impact and consistent downregulation following Fluv treatment, DEPDC1 was selected for further mechanistic investigation. We next examined the relationship between DEPDC1 expression and glycolysis-based risk stratification. DEPDC1 expression was significantly elevated in tumors from patients classified in the high glycolysis-risk group compared with those in the low-risk group (Fig. S2A). Moreover, DEPDC1 expression was markedly higher in LUAD tumor tissues than in adjacent normal tissues (Fig. S2B), further supporting its upregulation in LUAD.

To further elucidate the biological pathways associated with DEPDC1, LUAD patients in the TCGA cohort were stratified into high- and low-expression groups based on the median DEPDC1 level. Differential expression analysis followed by KEGG enrichment analysis showed that glycolysis-related signaling pathways were significantly enriched in the DEPDC1-high group (Fig. 2F), indicating a strong association between DEPDC1 expression and glycolysis. Correlation analysis using the TIMER database demonstrated a positive association between DEPDC1 and LDHA mRNA expression in LUAD samples (Fig. 2G). Consistently, correlation heatmap analysis showed that DEPDC1 (highlighted in red) was positively correlated with multiple glycolysis-associated genes (Fig. S2C), further supporting its involvement in glycolytic regulation. Violin plot analysis across tumor stages (Stage I-IV) revealed a progressive increase in DEPDC1 expression with advancing tumor stage (Fig. S2D), suggesting a potential role in LUAD progression. Given the observed link between DEPDC1 and glycolysis, we next examined its association with immune regulatory markers. Tumors with high DEPDC1 expression exhibited significantly elevated levels of immune checkpoint-related genes, including LAG3, PDCD1, CD274, PDCD1LG2 and IDO1 (Fig. S2E). Moreover, glycolysis-associated genes such as SLC2A1, HK2, PFKP and LDHA were markedly upregulated in the DEPDC1-high group (Fig. S2F). Together, these findings indicate that elevated DEPDC1 expression is associated with enhanced glycolysis and an immunosuppressive TME, supporting its role as a metabolic–immune regulatory factor in LUAD.

Fig. 2
Fig. 2
Full size image

DEPDC1 is associated with glycolysis in LUAD following Fluv treatment. (A) KEGG pathway enrichment analysis of DEGs between Fluv and Control groups. (B) Heatmap of intersecting genes among Fluv vs. Control DEGs, LUAD DEGs, and immune-related gene sets. (C) Heatmap of intersecting genes among Fluv vs. Control DEGs, LUAD DEGs, and glycolysis-related gene sets. (D) Relative mRNA expression of immune-related genes in A549 and H1299 LUAD cells treated with Fluv vs. Control. (E) Relative mRNA expression of glycolysis-related genes in A549 and H1299 LUAD cells treated with Fluv vs. Control. (F) KEGG pathway enrichment analysis of DEGs in TCGA-LUAD patients stratified by DEPDC1 expression (high vs. low). (G) Correlation analysis of DEPDC1 and LDHA mRNA expression in LUAD samples. (H) Bar plot showing differential metabolites between the Fluv and Control groups. PEP, Phosphoenolpyruvic acid; 2-PGA, 2-phosphoglyceric acid; PEA, phosphoethanolamine; cAMP, cyclic adenosine monophosphate. (I) Pathway enrichment analysis of differential metabolites between Fluv and Control groups

3. Fluvastatin regulates glycolysis in LUAD cells through DEPDC1

To determine whether Fluv alters metabolic reprogramming in LUAD cells, A549 cells were treated with Fluv or Control followed by targeted metabolomic profiling. PCA revealed a clear separation between the Fluv and Control groups, indicating substantial alterations in global metabolic signatures upon Fluv treatment (Fig. S2G). Differential metabolite analysis between the Fluv and Control groups was visualized using a heatmap (Fig. S2H). Among the significantly altered metabolites, several key glycolytic intermediates were affected, including Phosphoenolpyruvic acid (PEP), 2-phosphoglyceric acid (2-PGA), phosphoethanolamine (PEA), and cyclic adenosine monophosphate (cAMP). Specifically, PEP and 2-PGA levels were markedly decreased in the Fluv group compared with Control, whereas PEA and cAMP were increased (Fig. 2H), suggesting a potential reduction in glycolytic activity following Fluv treatment. To further characterize pathway-level changes, KEGG enrichment analysis of differentially expressed metabolites was performed. The results demonstrated significant enrichment in glycolysis/gluconeogenesis (Fig. 2I), as well as pyruvate metabolism, pentose phosphate pathway, carbon metabolism, and central carbon metabolism (Fig. S3A). Collectively, these findings indicate that Fluv treatment induces metabolic changes in LUAD cells, particularly affecting glycolytic and central carbon metabolic pathways.

Building on the transcriptional effects of Fluv and the TCGA analyses, we hypothesized that Fluv modulates glycolytic metabolism in A549 cells via DEPDC1. We first assessed the baseline expression of DEPDC1 across four LUAD cell lines (Fig. S2I) and found that A549 and H1299 cells exhibited relatively high DEPDC1 levels. These two cell lines were thus selected for further investigation of the relationship between DEPDC1 and glycolysis-related pathways. Overexpression of DEPDC1 in A549 and H1299 cells resulted in a marked increase in LDHA expression relative to Control group (Fig. S2J-K). Furthermore, DEPDC1 overexpression significantly upregulated key glycolytic protein, including GLUT1, HK2, and LDHA (Fig. 3A-B), suggesting that DEPDC1 may contribute to the regulation of glycolysis-related pathways in LUAD cells. To complement these findings, we evaluated four shRNA constructs targeting DEPDC1 in H1299 cells. Sh_DEPDC1_1 and Sh_DEPDC1_2 effectively suppressed DEPDC1 expression (Fig. S2L) and were used for subsequent knockdown experiments. DEPDC1 knockdown in both A549 and H1299 cells significantly downregulated the expression of GLUT1, HK2, and LDHA (Fig. 3C-F), further supporting a role for DEPDC1 in regulating glycolysis in LUAD cells.

4. DEPDC1 overexpression activates the PI3K-AKT-mTOR pathway and promotes LUAD cell proliferation

To investigate potential signaling mechanisms underlying DEPDC1-mediated glycolysis, TCGA-LUAD patients were stratified into high- and low-DEPDC1 expression groups and subjected to Gene Set Enrichment Analysis (GSEA). The analysis revealed significant enrichment of the PI3K-AKT-mTOR signaling pathway in the DEPDC1-high group (Fig. 3G; NES = 2.34, P < 0.001). Given that the PI3K-AKT-mTOR axis is a well-established upstream regulator of glycolysis [19], and previous studies have reported that DEPDC1 can enhance glycolytic flux through AKT/mTOR/HIF1α signaling in renal carcinoma [20], we hypothesized that DEPDC1 may mediate the effects of Fluv on glycolysis in LUAD through this pathway.

Additionally, targeted metabolomics revealed that Fluv may suppresses the expression of proteins associated with the PI3K-AKT-mTOR pathway (Fig. S4A-B). Consistently, overexpression of DEPDC1 in A549 and H1299 cells led to increased levels of PI3K, phosphorylated AKT/total AKT, and phosphorylated mTOR/total mTOR compared with Control group (Fig. 3H-I), confirming that DEPDC1 positively regulates PI3K-AKT-mTOR signaling. Conversely, DEPDC1 knockdown in both cell lines resulted in reduced expression of these pathway components (Fig. 3J-K). Functional assays further indicated that DEPDC1 promotes LUAD cell proliferation. In colony formation assays, DEPDC1-overexpressing A549 and H1299 cells formed significantly more colonies than Control group (Fig. S5A, Fig. S5C), whereas knockdown of DEPDC1 using two independent shRNAs in LLC, A549, and H1299 cells markedly reduced colony numbers (Fig. S5B, Fig. S5D-E).

Fig. 3
Fig. 3
Full size image

DEPDC1 is associated with glycolysis-related protein expression and PI3K-AKT-mTOR signaling in LUAD. (A) Expression of GLUT1, HK2, and LDHA in A549 cells overexpressing DEPDC1. (B) Expression of GLUT1, HK2, and LDHA in H1299 cells overexpressing DEPDC1. (C-D) Expression of GLUT1, HK2, and LDHA in A549 cells after DEPDC1 knockdown. (E-F) Expression of GLUT1, HK2, and LDHA in H1299 cells after DEPDC1 knockdown. (G) GSEA analysis of the PI3K-AKT-mTOR signaling pathway in DEPDC1 high- vs. low-expression groups in TCGA-LUAD. NES = 2.34, P < 0.001. (H) Expression of PI3K-AKT-mTOR pathway-related proteins in A549 cells overexpressing DEPDC1. (I) Expression of PI3K-AKT-mTOR pathway-related proteins in H1299 cells overexpressing DEPDC1. (J) Expression of PI3K-AKT-mTOR pathway-related proteins in A549 cells with DEPDC1 knockdown. (K) Expression of PI3K-AKT-mTOR pathway-related proteins in H1299 cells with DEPDC1 knockdown. (L) Expression levels of GLUT1, HK2, and LDHA proteins in LLC cells after Depdc1 knockdown. (M) Expression of PI3K-AKT-mTOR pathway-related proteins in LLC cells after Depdc1 knockdown

5. DEPDC1 knockdown suppresses LLC tumor growth and improves the tumor immune microenvironment

We first validated the knockdown efficiency of four Sh_Depdc1 constructs in LLC cells, finding that Sh_Depdc1_3 and Sh_Depdc1_4 effectively reduced DEPDC1 expression (Fig. S5F). DEPDC1 knockdown significantly decreased the expression of the glycolysis-related proteins GLUT1, HK2, and LDHA (Fig. 3L), as well as key components of the PI3K-AKT-mTOR pathway, including PI3K, phosphorylated AKT/total AKT, and phosphorylated mTOR/total mTOR (Fig. 3M). To assess the in vivo relevance, LLC cells transduced with Control, Sh_Depdc1_3, or Sh_Depdc1_4 were implanted subcutaneously into C57BL/6J mice. Tumor growth monitoring revealed that DEPDC1 knockdown markedly suppressed tumor growth rates (Fig. 4B), with Sh_Depdc1_3- and Sh_Depdc1_4-derived tumors exhibiting significantly lower tumor weights compared with Control group (Fig. 4A). Notably, there were no significant differences in body weight among the three groups (Fig. 4C), suggesting no apparent systemic toxicity. Subsequently, Immunohistochemical analysis of tumors from Control, Sh_Depdc1_3, and Sh_Depdc1_4 groups confirmed that DEPDC1 expression was markedly reduced in DEPDC1-knockdown LLC tumors (Fig. S5G), validating the successful establishment of the in vivo model. Correspondingly, LDHA expression was also significantly decreased in DEPDC1- knockdown tumors compared with Control group (Fig. S5H). Flow cytometry analysis of the TME revealed that DEPDC1 knockdown markedly increased infiltration of M1 macrophages (Fig. 4D) and effector T cells (Fig. 4F), while the proportion of immunosuppressive M2 macrophages was significantly reduced (Fig. 4E). Collectively, these findings suggest that DEPDC1 knockdown reduces DEPDC1 and LDHA expression in LUAD and promotes a more antitumor TIME.

Fig. 4
Fig. 4
Full size image

sh_DEPDC1 promotes antitumor immunity and Fluv + NPL exerts therapeutic effects in LUAD. (A) Antitumor effect of Depdc1 knockdown in LLC tumor-bearing mice. (B) Tumor growth curves of control and Depdc1-knockdown (n = 5). (C) Body weight changes in control and Depdc1-knockdown mice. (D) Percentage of M1 macrophages in LLC tumor tissues (n = 3). (E) Percentage of M2 macrophages in LLC tumor tissues (n = 3). (F) Percentage of effector T cells in LLC tumor tissues (n = 3). (G) Chemical structure of the TPA-DCR photosensitizer. (H) Photograph of the cRGD-TPA-DCR-NP solution. (I) Transmission electron microscopy (TEM) image of cRGD-TPA-DCR NP. (J) Particle size distribution of cRGD-TPA-DCR NP. (K) Cytotoxicity of different treatments in A549 cells. (L) Lactate levels in A549 cell supernatants. (M) Intracellular reduced glutathione (GSH) levels in A549 cells. (N) Relative extracellular ATP release in A549 cells (vs. Control)

6. cRGD-TPA-DCR-NP-mediated photodynamic therapy combined with Fluv enhances LUAD cells cytotoxicity and suppresses glycolysis

Previous studies have reported that photosensitizers or nanomedicines can modulate glycolytic metabolism in LUAD, thereby affecting tumor growth [12, 21, 22]. Based on this, we next investigated the combined effect of a photosensitizer and Fluv. Following our group’s prior experience [23], we synthesized the TPA-DCR photosensitizer (Fig. 4G). The chemical structure of TPA-DCR was confirmed by HRMS (Fig. S6C), ¹H NMR (Fig. S6A), and ¹³C NMR (Fig. S6B). HRMS (ESI⁺) analysis gave a calculated m/z [M + H]⁺ value of 497.14306 (C₃₁H₂₁N₄OS⁺), with the experimentally measured value of 497.14357. ¹H NMR (DMSO-d₆) spectra exhibited characteristic aromatic proton resonances in the range of δ 6.96–7.93 ppm as multiple overlapping multiplets, attributable to the triphenylamine and conjugated aromatic moieties. ¹³C NMR spectra showed characteristic carbon signals distributed between δ 109.67 and 166.26 ppm, including aromatic and olefinic carbons, electron-deficient sp² carbons at δ 165.87–166.26 ppm, and cyano carbon resonances around δ 113–114 ppm. These signals collectively agreed well with the expected molecular structure.

Flow cytometry analysis of Integrin αVβ3 (CD51⁺CD61⁺) expression in 3T3, LLC, A549, and H1299 cells (Fig. S6F) revealed low expression in normal 3T3 cells, whereas LLC, A549, and H1299 LUAD cells exhibited high expression. Based on this, we developed cRGD-functionalized TPA-DCR nanoparticles (cRGD-TPA-DCR-NP) targeting Integrin αVβ3 (CD51⁺CD61⁺), enabling precise delivery of the photosensitizer to LUAD cells for photodynamic therapy. We further performed cRGD conjugation validation. First, prior to characterization, unreacted free cRGDfKC peptides were thoroughly removed by dialysis and ultrafiltration to eliminate potential interference from residual free peptides. Subsequently, high-resolution mass spectrometry (HRMS) was performed on the purified cRGD-NP samples to verify the presence of cRGDfKC in the nanoparticles. Free cRGDfKC peptide exhibited a characteristic ion peak at m/z 705.32, while a corresponding peak was still detected at m/z 705.31 in the purified cRGD-NP samples (Fig. S7A-B). Since unbound free peptides had been removed before analysis, the detection of this characteristic signal indicates that cRGDfKC was successfully conjugated to the nanoparticles.

The cRGD-TPA-DCR-NPs were prepared via the nanoprecipitation method (Fig. 4H-I) and exhibited an average particle size of approximately 93 nm (Fig. 4J). The encapsulation efficiency of TPA-DCR in cRGD-TPA-DCR-NPs was 91.6%, and the drug loading efficiency was 23.39%. Spectroscopic analysis of cRGD-TPA-DCR-NP showed a maximum absorption wavelength at 452 nm (Fig. S6D) and maximum emission at 616 nm (Fig. S6E). While these optical properties are suitable for in vitro characterization, fluorescence signals in the visible spectral range are generally not ideal for quantitative in vivo circulation half-life and biodistribution studies because they are more susceptible to tissue autofluorescence, hemoglobin absorption, and photon scattering. To achieve reliable in vivo tracking, we adopted a widely used strategy in nanomedicine by covalently conjugating Cy5 to the nanoparticle surface [24,25,26]. Compared with the intrinsic fluorescence of cRGD-TPA-DCR-NP, Cy5 is more suitable for in vivo imaging due to its improved signal-to-background characteristics. Therefore, Cy5 was conjugated to the surface of cRGD-TPA-DCR-NP or TPA-DCR-NP and used as a surrogate imaging probe to track the nanoparticles in vivo.

To assess the blood circulation profile of Cy5-cRGD-TPA-DCR-NP (Cy5-cRGD-NP), Cy5-cRGD-NP was intravenously injected into C57BL/6 mice. The results showed that Cy5-cRGD-NP exhibited a relative circulation half-life of approximately 6.66 h, suggesting relatively prolonged blood circulation in vivo (Fig. S7E). To further evaluate biodistribution of Cy5-cRGD-NP and Cy5-NP, liver, spleen, and tumor tissues were collected 4 h after intravenous administration. Compared with the non-targeted Cy5-NP, targeted Cy5-cRGD-NP exhibited significantly higher fluorescence intensity in LLC tumor tissues (Fig. S7F), indicating that cRGD modification can enhance the LLC tumor-targeting capability of nanoparticles to a certain extent.

In addition, the stability of cRGD-TPA-DCR-NP was evaluated. The results showed that the average particle size remained stable at approximately 93 nm (Fig. S7C), while the PDI remained around 0.17 (Fig. S7D) throughout the observation period, with no obvious aggregation or size increase detected. These findings indicate good stability of the nanoparticles. Finally, we compared the antitumor efficacy of targeted cRGD-TPA-DCR-NP and non-targeted TPA-DCR-NP under light irradiation in LLC tumor-bearing mice (Fig. S7G). The results demonstrated that cRGD-modified nanoparticles exhibited significantly greater tumor growth inhibition than the non-targeted nanoparticles.

CCK-8 assays demonstrated that Fluv reduced the viability of A549 (Fig. S7L) and H1299 (Fig. S7M) cells in a dose-dependent manner, indicating its cytotoxic effect against LUAD cells. Subsequently, we evaluated the potential cytotoxicity of 200 µM fluvastatin in several normal cell lines, including NIH-3T3 fibroblasts, L929 fibroblasts, HUVEC endothelial cells, and HEK293T cells (Fig. S7J). The results showed that treatment with 200 µM fluvastatin did not induce significant cytotoxicity in these normal cell lines compared with the control group. The IC50 of Fluv for both A549 and H1299 cells was approximately 200 µM, this concentration falls within the effective concentration range for both cell lines and was able to induce a pronounced downregulation of DEPDC1 expression (Fig. S7H-I). Therefore, 200 µM Fluv was used in subsequent anti-LUAD cell experiments [27, 28]. Treatment with NPL (cRGD-TPA-DCR-NP under white light, 25 mW/cm²) also inhibited A549 (Fig. S7N) and H1299 (Fig. S7O) cell viability in a concentration-dependent manner, confirming a pronounced photodynamic effect.

The combination of Fluv and NPL further enhanced cytotoxicity against A549 (Fig. 4K), H1299 (Fig. S8A), and LLC cells (Fig. S8B), indicating a synergistic effect on LUAD cells. Treatment with 200 µM Fluv + 12.5 µM NPL inhibited over 50% of A549 and H1299 cell viability, supporting the use of this combination in subsequent experiments. Notably, Fluv and NPL exhibited minimal toxicity toward normal HUVEC cells (Fig. S7K), suggesting good safety for non-tumor cells. Fluv + NPL co-treatment significantly reduced extracellular lactate levels in A549 (Fig. 4L) and H1299 cells (Fig. S8C) compared with either monotherapy. Fluv treatment has been reported to weaken the antioxidant defenses (GSH reserves) in certain tumor types [29,30,31], which was further amplified by NPL-mediated hydroxyl radical “burst” [23], leading to a more pronounced decrease in GSH levels compared with either monotherapy (Fig. 4M, Fig. S8D). We further assessed extracellular ATP release in the culture supernatants of A549 (Fig. 4N) and H1299 (Fig. S8E) cells subjected to different treatments. The results showed that the Fluv + NPL group exhibited significantly higher extracellular ATP levels than the Control, Fluv, and NPL groups. As ATP is one of the important damage-associated molecular patterns (DAMPs) released during immunogenic cell death (ICD), its release can promote the recruitment, maturation, and antigen presentation of dendritic cells (DCs) [32, 33].

We further investigated the effects of Fluv + NPL on the PI3K-AKT-mTOR signaling pathway and glycolysis in A549 and H1299 cells. Western blot analysis revealed that, compared with the Control and Fluv groups, Fluv + NPL treatment reduced the expression of the glycolysis-related proteins GLUT1, HK2, and LDHA (Fig. 5A-B). Moreover, Fluv + NPL more effectively inhibited PI3K expression as well as the phosphorylation of AKT and mTOR than either the Control or Fluv treatment alone (Fig. 5C-D). In addition, Seahorse metabolic flux analyses (GlycoPER and OCR) were performed in both A549 and H1299 cells (Fig. 5E-K, S8F-L). The results showed that DEPDC1 knockdown significantly suppressed glycolytic activity, whereas DEPDC1 overexpression partially restored the glycolysis inhibited by Fluv + NPL treatment, further supporting the involvement of DEPDC1 in metabolic regulation. Consistent with these findings, lactate level was significantly reduced in the culture supernatants of both A549 (Fig. 5L) and H1299 (Fig. 5M) cells following Fluv + NPL treatment compared with the Control group. Notably, DEPDC1 overexpression largely reversed lactate level. Collectively, these findings suggest that Fluv + NPL suppresses PI3K-AKT-mTOR signaling and glycolysis in LUAD cells at least in part through downregulation of DEPDC1.

Fig. 5
Fig. 5
Full size image

Fluv + NPL suppresses DEPDC1-associated glycolysis in LUAD. (A) Expression levels of GLUT1, HK2, and LDHA in A549 cells of Control, Fluv, and Fluv + NPL groups. (B) Expression levels of GLUT1, HK2, and LDHA in H1299 cells of Control, Fluv, and Fluv + NPL groups. (C) Expression of DEPDC1 and PI3K-AKT-mTOR pathway-related proteins in A549 cells of Control, Fluv, and Fluv + NPL groups. (D) Expression of DEPDC1 and PI3K-AKT-mTOR pathway-related proteins in H1299 cells of Control, Fluv, and Fluv + NPL groups. (E-G) Glycolytic function of A549 cells under the indicated treatments (Control, Fluv + NPL, shDEPDC1, or Fluv + NPL + DEPDC1-OE), including GlycoPER (E), basal glycolysis (F), and compensatory glycolysis (G). (H-K) Mitochondrial respiration of A549 cells under the indicated treatments, including oxygen consumption rate (OCR) (H), basal respiration (I), maximal respiration (J), and ATP production (K). (L-M) Lactate levels in the culture supernatant of A549 (L) and H1299 (M) cells

7. Fluv + NPL promotes ROS generation, mitochondrial membrane potential loss, apoptosis, and ICD while reducing glucose uptake in LUAD cells

Treatment with Fluv + NPL markedly increased intracellular reactive oxygen species (ROS) levels in A549 (Fig. 6A) and H1299 (Fig. S9A) cells, as indicated by the strongest DCFH-DA green fluorescence compared with PBS, NP, Light, NPL, Fluv, and Fluv + NP groups. This suggests that the combination significantly enhances oxidative stress, a key driver of cell death. In TMRE staining assays, Fluv + NPL-treated A549 (Fig. 6B) and H1299 (Fig. S9B) cells exhibited loss of mitochondrial membrane potential, with markedly decreased red fluorescence intensity, indicating pronounced mitochondrial dysfunction closely associated with apoptosis. Furthermore, Fluv + NPL induced substantial apoptosis in A549 (Fig. 6C) and H1299 (Fig. S9C) cells, characterized by increased PI-positive (red, dead) staining and reduced Calcein-AM-positive (green, live) staining. Quantitative analysis confirmed that the apoptosis rate in the Fluv + NPL group was higher than in Fluv, NPL, or other Control groups, demonstrating a synergistic cytotoxic effect between Fluv and NPL.

Further immunofluorescence analysis revealed that Fluv + NPL treatment induced pronounced translocation of High Mobility Group Box 1 (HMGB1) from the nucleus to the cytoplasm/extracellular space in A549 (Fig. 6D) and H1299 (Fig. S9D) cells, as indicated by the weakest residual nuclear red fluorescence and the lowest relative intensity, suggesting a strong induction of ICD. Concurrently, Fluv + NPL-treated A549 (Fig. 6E) and H1299 (Fig. S9E) cells exhibited increased surface exposure of calreticulin (CRT), with the strongest green fluorescence and clear membrane localization, significantly higher than other treatment groups, further confirming potent ICD induction. Moreover, compared with PBS, Fluv, or NPL alone, Fluv + NPL significantly reduced uptake of the glucose analog 2-NBDG in A549 (Fig. 6F) and H1299 (Fig. S9F) cells, as evidenced by markedly diminished green fluorescence, indicating effective inhibition of glucose uptake. Collectively, Fluv + NPL exhibits potent synergistic anti-LUAD effects in A549 and H1299 cells, involving enhanced ROS production, loss of mitochondrial membrane potential, apoptosis, and ICD (as indicated by CRT surface exposure and HMGB1 release), accompanied by substantial inhibition of glucose uptake.

Fig. 6
Fig. 6
Full size image

Fluv + NPL treatment induces ROS generation, mitochondrial membrane potential disruption, apoptosis, ICD, and inhibits glucose uptake in A549 cells. (A) ROS detection in A549 cells using DCFH-DA staining. (B) Mitochondrial membrane potential assessment in A549 cells using TMRE staining. (C) Live/dead staining of A549 cells using Calcein AM and PI. (D) HMGB1 immunostaining in A549 cells. (E) CRT immunostaining in A549 cells. (F) Glucose uptake in A549 cells measured by 2-NBDG

8. Fluv + NPL reduces LUAD-derived exosomal CCL2 in association with DEPDC1 downregulation

Accumulating evidence indicates that tumor-derived exosomes can modulate the TME, thereby affecting the efficacy of immunotherapy [34,35,36,37]. However, the mechanisms by which exosomes regulate the TME remain largely unexplored. To investigate the effects of Fluv + NPL on LUAD cell-derived exosomes and their role in modulating the TME, we performed proteomic analysis of exosomes secreted by A549 cells under Control or Fluv + NPL treatment [38]. Scanning electron microscopy revealed that the exosomes exhibited typical spherical vesicular morphology (Fig. 7A), while nanoparticle tracking analysis (NTA) showed that the average diameter of exosomes in all groups was approximately 137 nm (Fig. 7A). Western blot analysis confirmed the presence of the exosomal markers Alix, TSG101, and CD63 in isolated exosomes (Fig. 7B). Furthermore, Dil-labeled A549-derived exosomes were internalized by macrophages (Fig. 7C), demonstrating that these exosomes are capable of intercellular transfer.

PCA of the exosomal proteomic data revealed a clear separation between the Fluv + NPL group and the Control group (Fig. S10A), with Fluv + NPL samples forming an independent cluster, indicating that this combined treatment profoundly alters the overall exosomal proteome. Consistent with the heatmap and volcano plot (Fig. 7D, Fig. S10B), the chemokine CCL2 was markedly downregulated in Fluv + NPL-treated exosomes. Venn analysis (Fig. 7E) identified CCL2 and PLAU as core overlapping candidates among differentially expressed exosomal proteins, LUAD differentially expressed genes (LUAD-DEGs), and immune-related genes (IRGs). Protein–protein interaction (PPI) network analysis further suggested that CCL2 represents an important node within the immune regulatory network (Fig. 7F). Western blot validation aligned closely with the proteomic data, showing that CCL2 protein levels in A549-derived exosomes from the Fluv + NPL group were significantly lower than those in the Control and Fluv-only groups (Fig. 7G). Interestingly, TIMER2.0 analysis revealed a positive correlation between DEPDC1 and CCL2 expression (Fig. S10C). In DEPDC1-overexpressing (DEPDC1_OE) A549 cells, the Fluv + NPL-induced downregulation of exosomal CCL2 was largely reversed (Fig. 7H), suggesting that DEPDC1 positively regulates exosomal CCL2 levels, and that its overexpression can counteract the inhibitory effect of Fluv + NPL, indicating a role for DEPDC1 in exosomal CCL2 regulation.

Fig. 7
Fig. 7
Full size image

Fluv + NPL regulates exosomal CCL2 via DEPDC1 to modulate macrophage polarization in LUAD. (A) Scanning electron microscopy (SEM) images and nanoparticle tracking analysis (NTA) profiles of exosomes derived from A549 cells. (B) Western blot analysis of exosomal markers in A549 cells. (C) Uptake of DiL-labeled A549-derived exosomes by macrophages. (D) Heatmap of differentially expressed exosomal proteins. (E) Intersection analysis of genes corresponding to differentially expressed exosomal proteins with LUAD DEGs and IRGs to identify candidate exosomal proteins associated with LUAD and immunity. (F) PPI network of differentially expressed exosomal proteins. (G) CCL2 expression levels in A549-derived exosomes. (H) DEPDC1 overexpression in A549 cells reverses Fluv + NPL-induced changes in exosomal CCL2 levels. (I) Correlation analysis between CCL2 expression and M2 macrophage infiltration in LUAD patients from the TCGA-LUAD cohort. (J) Heatmap of transcriptomic data from macrophages co-cultured with Fluv + NPL or Control exosomes

9. Fluv + NPL-derived exosomes regulate macrophage polarization via the DEPDC1–CCL2 axis

TIMER database analysis revealed a significant positive correlation between CCL2 expression and M2 macrophage infiltration in the LUAD TME (Fig. 7I). Consistently, TCGA analysis showed that patients with high CCL2 expression exhibited markedly increased M2 macrophage infiltration (Fig. S10D), suggesting that CCL2 may influence macrophage polarization. To investigate the effects of exosomes from different treatment groups on macrophage phenotype, we purified Control-exo, Fluv-exo, and Fluv + NPL-exo and co-cultured them with PMA-differentiated THP-1 macrophages. Subsequent pairwise differential gene analysis was used to construct PPI networks [39]for each comparison (Fig. S10E-G). CSF1 was identified across all comparisons, suggesting a role in exosome-mediated regulation of macrophage phenotype. Volcano plots and heatmaps further visualized differential gene expression, showing that, compared with Control-exo, both Fluv-exo (Fig. S10I, Fig. S11A) and Fluv + NPL-exo (Fig. S10H, Fig. 7J) significantly downregulated CSF1 expression in macrophages. Moreover, direct comparison between Fluv + NPL-exo and Fluv-exo demonstrated that Fluv + NPL-exo exerted a more pronounced inhibitory effect on CSF1 expression (Fig. S10J, Fig. S11B).

Subsequently, KEGG pathway enrichment analyses were performed on the DEGs among the three groups. KEGG pathway enrichment analysis revealed that DEGs between Fluv + NPL-exo and Control-exo were significantly enriched in immune-related pathways, including JAK-STAT signaling pathway, C-type lectin receptor signaling pathway, Cytokine-cytokine receptor interaction and HIF-1 signaling pathway (Fig. 8A). For Fluv-exo versus Control-exo, DEGs were mainly enriched in Chemokine signaling pathway, Cytokine-cytokine receptor interaction and PPAR signaling pathway (Fig. S11C). DEGs between Fluv + NPL-exo and Fluv-exo were significantly enriched in Hippo signaling pathway, FoxO signaling pathway, and PPAR signaling pathway (Fig. S11D). Taken together, KEGG enrichment analyses indicate that co-culture of macrophages with differently treated exosomes predominantly affects genes involved in cytokine-mediated signaling as well as immune-related pathways. These findings suggest that exosomes may modulate macrophage functional states by regulating cytokine-mediated signaling networks.

We further evaluated the different treatment groups using GSEA. Compared with Fluv + NPL-exo and Fluv-exo groups (Fig. S12A-B), macrophages co-cultured with Control-exo were significantly enriched in the TGF-β signaling pathway. This pathway is known to promote M2 macrophage polarization [40], inducing a pro-tumor phenotype characterized by enhanced immunosuppression and tumor angiogenesis. In contrast, compared with Fluv-exo (Fig. S12C-D), the Fluv + NPL-exo group was significantly enriched in interferon-α and interferon-γ response pathways. These pathways play important roles in antitumor immunity by promoting recruitment of T cells and NK cells, enhancing antitumor immune responses, and modulating PD-L1 expression to improve immune checkpoint blockade efficacy [41].

We further validated the transcriptomic findings at the cellular level in vitro. Macrophages were co-cultured with exosomes from different treatment groups, and the expression of macrophage-related marker genes was assessed by qPCR. Compared with the Control-exo group, treatment with Fluv-exo and Fluv + NPL-exo significantly upregulated the mRNA levels of M1 macrophage markers, including IL1β, iNOS, CD80, and CD86, with the Fluv + NPL-exo group showing a more pronounced effect (Fig. 8B). Conversely, M2 macrophage-associated markers, including CD163, CD206, IL10, and ARG1, were significantly downregulated in the Fluv-exo and Fluv + NPL-exo groups, with the strongest suppression observed in the Fluv + NPL-exo group (Fig. 8C). To further elucidate the role of DEPDC1 in this regulatory process, DEPDC1 was over-expressed in tumor cells, and exosomes were collected from Control, Fluv + NPL, and Fluv + NPL + DEPDC1_OE groups for macrophage co-culture. qPCR analysis showed that Fluv + NPL-exo significantly upregulated M1 markers (IL1β, iNOS, CD80, CD86) (Fig. 8D) and downregulated M2 markers (CD163, CD206, IL10, ARG1) (Fig. 8E) compared with Control-exo. Importantly, exosomes from the Fluv + NPL + DEPDC1_OE group reversed these expression trends, suggesting that DEPDC1 overexpression attenuates the Fluv + NPL-induced changes in macrophage polarization (Fig. 8D–E).

We further assessed the expression of surface markers on macrophages by flow cytometry following co-culture with exosomes. Compared with the Control-exo group, treatment with Fluv-exo and Fluv + NPL-exo significantly increased surface CD80 expression, with the Fluv + NPL-exo group showing the most pronounced upregulation (Fig. 8F). Conversely, CD206 expression was significantly reduced in the Fluv-exo and Fluv + NPL-exo groups, with the strongest decrease observed in the Fluv + NPL-exo group (Fig. 8G). Based on the analyses in Fig. 7I and Fig. S10D, CCL2 was suggested to influence macrophage polarization. To confirm the role of exosomal CCL2 in regulating macrophage phenotype, we performed rescue experiments by adding exogenous CCL2 to the co-culture system. The results showed that the upregulation of CD80 induced by Fluv + NPL-exo was attenuated by CCL2 addition (Fig. 8H), and the downregulation of CD206 was partially reversed (Fig. 8I), indicating that CCL2 mediates the modulatory effects of exosomes on macrophage polarization.

Subsequently, we performed additional experiments involving CCL2 knockdown and CCR2 functional blockade. Specifically, DEPDC1-overexpressing A549 cells were established, followed by CCL2 knockdown in donor tumor cells. The results showed that exosomes derived from DEPDC1-overexpressing cells promoted macrophage polarization toward the M2 phenotype. However, when CCL2 was knocked down in donor cells, this pro-M2 polarization effect was attenuated. In addition, treatment with the CCR2 inhibitor RS 504,393 in the co-culture system reversed the M2-polarizing effect induced by exosomes from DEPDC1-overexpressing cells (Fig. 8J, Fig. S10K). Notably, consistent results were also observed with exosomes derived from H1299 cells (Fig. 8K, Fig. S10L). These findings indicate that both knockdown of CCL2 in donor tumor cells and blockade of CCR2 signaling in recipient macrophages can significantly weaken the M2 polarization induced by exosomes from DEPDC1-overexpressing tumor cells, thereby supporting an important functional role for the CCL2/CCR2 signaling axis in this process.

In addition, we performed additional experiments using a murine cell system to determine whether the DEPDC1–CCL2 axis is involved in LLC-derived exosome-mediated macrophage polarization. Three experimental groups were established, including Control, sh_Depdc1, and sh_Depdc1 + CCL2, and exosomes were isolated from LLC cells in each group. Subsequently, exosomes from the different treatment groups were co-cultured with bone marrow-derived macrophages (BMDMs) differentiated in vitro from bone marrow cells isolated from C57BL/6J mice. These results (Fig. S14D-E) showed that, compared with exosomes derived from the Control group, exosomes derived from sh_Depdc1-treated LLC cells significantly suppressed M2 polarization while promoting M1 polarization of BMDMs. Notably, the addition of CCL2 to BMDMs treated with sh_Depdc1-derived exosomes reversed these effects. These findings indicate that DEPDC1 knockdown–associated exosomal alterations in murine LLC cells are associated with macrophage polarization, which may involve CCL2-related signaling pathways.

Fig. 8
Fig. 8
Full size image

Fluv + NPL modulates macrophage polarization via exosomal CCL2 in LUAD. (A) KEGG pathway enrichment analysis of differentially expressed genes between Fluv + NPL and Control exosome-treated macrophages. (B) M1 marker mRNA levels in exosome-treated macrophages. (C) M2 marker mRNA levels in exosome-treated macrophages. (D) Exosomes derived from DEPDC1-overexpressing A549 cells under Fluv + NPL treatment inhibit M1 macrophage marker mRNA expression. (E) Exosomes derived from DEPDC1-overexpressing A549 cells under Fluv + NPL treatment reverse M2 macrophage marker mRNA expression. (F) Proportion of M1 macrophages after treatment with Control, Fluv, or Fluv + NPL exosomes. (G) Proportion of M2 macrophages after treatment with Control, Fluv, or Fluv + NPL exosomes. (H) Proportion of M1 macrophages in Control, Fluv + NPL, and Fluv + NPL+ CCL2 treatment groups. (I) Proportion of M2 macrophages in Control, Fluv + NPL, and Fluv + NPL+ CCL2 treatment groups. (J-K) Exosomes from A549 (J) and H1299 (K) cells with DEPDC1-OE or shCCL2 were co-cultured with macrophages, with or without the CCR2 inhibitor RS504393. Polarization-related proteins were assessed by Western blot

10. Fluv + NPL inhibits LUAD growth and modulates the tumor immune microenvironment

Based on the above in vitro results, Fluv + NPL combination treatment not only suppressed glycolysis-related signaling in LUAD cells but also reduced exosomal CCL2 levels, which may contribute to changes in macrophage polarization and potentially enhance antitumor immune responses. We next evaluated whether the cRGD-NP could efficiently target LUAD in mice and whether the immune-metabolic regulatory effects observed in vitro could be recapitulated in vivo. Accordingly, we evaluated the in vivo biodistribution of cRGD-NP, assessed the anti-tumor efficacy of Fluv + NPL combination therapy, and examined its overall impact on the TIME in a mouse LUAD model.

Given that the TPA-DCR photosensitizer exhibits insufficient fluorescence in the near-infrared I (NIR-I) region (its absorption/emission primarily lies within the visible-red light range), we substituted it with the lipophilic NIR-I fluorophore Cy7-NHS for encapsulation, enabling evaluation of the in vivo biodistribution of cRGD-NP. Biodistribution studies via tail vein injection (Fig. 9A) showed that free Cy7-NHS primarily accumulated in the liver, indicating its poor tumor-targeting capability. In contrast, both NP-encapsulated Cy7-NHS and cRGD-NP-encapsulated Cy7-NHS exhibited enhanced tumor accumulation. Notably, the cRGD-NP group demonstrated the strongest tumor-specific fluorescence signal from 4 to 12 h post-injection, suggesting that αVβ3 integrin-mediated active targeting significantly improves LLC tumor enrichment of nanoparticles.

The in vivo treatment regimen for the LLC tumor model is shown in Fig. 9B. Compared with the Fluv + NP, NPL, Fluv, Light, NP, and PBS groups, the Fluv + NPL group exhibited the slowest LLC tumor growth (Fig. 9C) and tumor weight (Fig. S13B) significantly reduced, demonstrating the strongest anti-tumor efficacy. Body weight changes across all treatment groups showed no significant differences (Fig. S13A), indicating good tolerability. Immunohistochemical analysis of tumor tissues from the PBS, Fluv, and Fluv + NPL groups revealed that DEPDC1 (Fig. 9G) and LDHA (Fig. S13F) levels were lowest in the Fluv + NPL group, suggesting that the combination treatment effectively suppresses DEPDC1 and LDHA expression, which may contribute to reduced glycolytic activity in tumors. Furthermore, immunofluorescence staining revealed that CRT exposure on the tumor cell surface was significantly increased, whereas HMGB1 expression was reduced in Fluv + NPL-treated tumors (Fig. 9H), compared with the Fluv and PBS groups, indicating that enhanced immunogenic cell death may promote greater infiltration of immune cells into LLC tumor.

Next, LLC tumor tissues from the mice were analyzed by flow cytometry and immunofluorescence. Compared with the Fluv and PBS groups, Fluv + NPL treatment significantly increased the infiltration of M1 macrophages (Fig. 9D, Fig. S13C, Fig. 9I) and effector T cells (Fig. 9F, Fig. S13E) in the TME, while markedly reducing the proportion of M2 macrophages (Fig. 9E, Fig. S13D, Fig. 9I). Additionally, H&E staining of major organs (heart, liver, spleen, lung, and kidney) revealed no obvious pathological abnormalities or organ damage (Fig. S13G). Hematological parameters, including red blood cells, white blood cells, platelets, hematocrit, hemoglobin, and mean corpuscular hemoglobin (Fig. S13H-M), as well as serum biochemical markers (ALT, AST, creatinine, and urea) (Fig. S13N-Q), remained within normal ranges and showed no significant differences compared with the PBS group, indicating good biocompatibility of all treatment regimens.

To further evaluate the immunomodulatory effects of Fluv + NPL and DEPDC1 in LUAD and to assess the generalizability of our findings, we established a syngeneic CMT167 tumor model in immunocompetent C57BL/6J mice (Fig. 9J, S14F-G). Three groups were included: Control, Fluv + NPL, and Fluv + NPL+DEPDC1_OE. Flow cytometric analysis revealed that, compared with the Control group, Fluv + NPL treatment modulated the TIME, as evidenced by a robust increase in M1-like macrophages and a decrease in M2-like macrophages (Fig. 9K-L, S14H-I). In addition, dendritic cell (DC) activation was significantly enhanced (Fig. 9M, S14J), accompanied by increased infiltration of cytotoxic CD8⁺ T cells (CD8⁺GZMB⁺IFN-γ⁺) (Fig. 9N, S14K). Notably, these effects were attenuated by DEPDC1 overexpression (Fluv + NPL + DEPDC1_OE), suggesting that the immunomodulatory activity of Fluv + NPL is at least partially mediated by DEPDC1 downregulation, but whether these immune alterations are functionally required for its antitumor efficacy remains to be determined.

To determine whether these immune alterations are functionally required for the therapeutic efficacy of Fluv + NPL, we performed in vivo immune cell depletion studies (Fig. S14A-C). Specifically, mice were treated with Fluv + NPL alone, Fluv + NPL combined with macrophage depletion (Fluv + NPL+Clodronate), or Fluv + NPL combined with both macrophage and CD8⁺ T-cell depletion (Fluv + NPL+Clodronate + αCD8). The results showed that either macrophage depletion alone or the combined depletion of macrophages and CD8⁺ T cells significantly impaired the antitumor efficacy of Fluv + NPL, as reflected by markedly reduced tumor growth inhibition. These findings suggest that the antitumor activity of Fluv + NPL may depend, at least in part, on the coordinated participation of macrophages and CD8⁺ T cells.

Fig. 9
Fig. 9
Full size image

Fluv + NPL enhance anti-tumor therapeutic efficacy through DEPDC1-associated regulation in LUAD. (A) In vivo near-infrared I (NIR-I) fluorescence imaging of free drugs and various nanoparticles in mice. (B) Schematic diagram of the LUAD mouse treatment regimen. (C) Tumor growth curves of LLC tumor-bearing mice (n = 5). (D) Percentage of M1 macrophages in LLC tumor tissues (n = 3). (E) Percentage of M2 macrophages in LLC tumor tissues (n = 3). (F) Percentage of effector T cells in LLC tumor tissues (n = 3). (G) Immunohistochemical analysis of DEPDC1 expression in LLC tumor tissues (n = 3). (H) Immunofluorescence staining of HMGB1 and CRT in LLC tumor tissues (n = 3). (I) Immunofluorescence staining of M1 and M2 macrophages in LLC tumor tissues (n = 3). (J) Tumor growth curves of CMT167 tumor-bearing mice (n = 5). (K) Percentage of M1 macrophages in CMT167 tumor tissues (n = 5). (L) Percentage of M2 macrophages in CMT167 tumor tissues (n = 5). (M) Percentage of mature dendritic cells (mDCs) in draining lymph nodes (n = 5). (N) Percentage of cytotoxic CD8+ T cells in CMT167 tumor tissues (n = 5)

11. DEPDC1 activates NF-κB signaling via ZNF224 to regulate PI3K-AKT and CCL2 expression

These results establish a functional link between DEPDC1 regulation and antitumor immunity. However, the downstream signaling pathways through which DEPDC1 exerts these effects remain to be defined. Previous studies [42, 43] have reported that DEPDC1 interacts with ZNF224 and activates downstream NF-κB signaling. Based on these findings, we investigated whether DEPDC1 regulates glycolysis and immunosuppressive signaling through the ZNF224/NF-κB axis. First, we performed reciprocal co-immunoprecipitation (Co-IP) assays in A549 and H1299 cells. The results consistently demonstrated interaction between DEPDC1 and ZNF224 in both cell lines (Fig. 10A-B). Second, nuclear-cytoplasmic fractionation and immunofluorescence analyses revealed that DEPDC1 overexpression significantly promoted nuclear translocation of NF-κB p65 in A549 cells (Fig. 10C-D). Similar results were observed in H1299 cells, indicating that DEPDC1 can activate NF-κB signaling (Fig. S15A-B). Third, rescue experiments using the NF-κB inhibitor BAY 11-7082 were conducted. Western blot analysis showed that DEPDC1 overexpression markedly increased NF-κB p65 and IκBα activation, whereas BAY 11-7082 substantially attenuated these effects (Fig. 10F, S15D). Importantly, previous studies have demonstrated that NF-κB directly regulates the transcription of both PIK3CA [44, 45]and CCL2 [46, 47]. Therefore, we further examined whether NF-κB activation contributes to DEPDC1-associated regulation of PI3K signaling and CCL2 expression. qPCR and WB analysis analysis consistently showed that DEPDC1 overexpression significantly increased the expression of PI3K(PIK3CA) and CCL2, whereas inhibition of NF-κB largely reversed these effects (Fig. 10E and G, S15C, S15E).

Furthermore, we performed DEPDC1 rescue studies under Fluv + NPL treatment (Fig. 10H-I, S15F-G). Western blot analysis demonstrated that Fluv + NPL markedly suppressed NF-κB activation, PI3K-AKT signaling, and CCL2 expression. Importantly, DEPDC1 overexpression significantly attenuated these inhibitory effects. Collectively, these findings suggest that DEPDC1 contributes to NF-κB-associated regulation of PI3K–AKT signaling and CCL2 expression, and that these effects are attenuated by Fluv + NPL treatment.

Fig. 10
Fig. 10
Full size image

DEPDC1 activates NF-κB signaling to regulate CCL2 and PI3K/AKT pathways. (A-B) Interaction between DEPDC1 and ZNF224 in A549 (A) and H1299 (B) cells was validated by Co-IP. (C) Nuclear and cytoplasmic fractionation followed by Western blot analysis in DEPDC1-overexpressing A549 cells. (D) Nuclear translocation of NF-κB p65 in control and DEPDC1-overexpressing A549 cells detected by immunofluorescence. (E) CCL2 and PIK3CA mRNA expression following DEPDC1 overexpression with or without the NF-κB inhibitor BAY 11-7082 in A549 cells. (F) NF-κB signaling protein levels following DEPDC1 overexpression with or without BAY 11-7082 in A549 cells. (G) PI3K/AKT signaling protein and CCL2 levels following DEPDC1 overexpression with or without BAY 11-7082 in A549 cells. (H) NF-κB signaling protein levels following Fluv + NPL treatment with or without DEPDC1 overexpression in A549 cells. (I) PI3K/AKT signaling protein and CCL2 levels following Fluv + NPL treatment with or without DEPDC1 overexpression in A549 cells

Discussion

In this study, we elucidated the role of DEPDC1 in glycolytic reprogramming and tumor-immune regulation in LUAD, and developed a glycolysis-targeted nanophotodynamic combination strategy. Our major findings are as follows: (1) a glycolysis-related gene-based risk score reliably predicts prognosis in LUAD patients and is associated with an immunosuppressive TME; (2) DEPDC1 is associated with glycolytic metabolism and immune modulation; (3) mechanistically, DEPDC1 interacts with ZNF224 to activate NF-κB signaling, thereby promoting PI3K–AKT–mTOR pathway activation and glycolytic metabolism; (4) in parallel, DEPDC1 regulates exosomal CCL2 levels, contributing to M2 macrophage polarization; and (5) targeting DEPDC1 with Fluv + NPL effectively suppresses glycolysis, improves the TIME, and inhibits lung adenocarcinoma growth.

Enhanced glycolysis is a hallmark of metabolic reprogramming during the initiation and progression of LUAD [48]. Our glycolysis-based prognostic risk score model demonstrated robust predictive performance in both training and internal validation cohorts of LUAD patients, and was further validated in independent external GEO datasets. The model remained stable across diverse clinical subgroups, indicating that glycolytic phenotypes not only reflect metabolic activity but also serve as reliable and generalizable predictors of clinical prognosis. Moreover, patients in the high-risk group exhibited elevated TMB and reduced ImmuneScore/ESTIMATEScore, suggesting a close association between enhanced glycolysis and immune evasion. These findings align with previous reports showing that lactate accumulation and TME acidification are associated with immunosuppressive remodeling, further highlighting the dual role of glycolysis as a metabolic and immunoregulatory mediator in tumors [49, 50].

Multi-omics analyses revealed that Fluv treatment markedly altered the transcriptomic and metabolic landscape of LUAD cells, with particularly pronounced effects on glycolysis and gluconeogenesis pathways. By intersecting differentially expressed genes with glycolysis- and immune-related gene sets, we identified DEPDC1 as a candidate molecule associated with both prognosis and glycolytic activity. Subsequent database analyses and validation in clinical samples demonstrated that tumors with high DEPDC1 expression exhibited not only upregulation of glycolytic enzymes but also elevated expression of immune checkpoint molecules, suggesting that DEPDC1 may be associated with both glycolytic metabolism and immune suppression.

This study presents a combinatorial therapeutic strategy integrating targeted photosensitizer-loaded nanoparticles (cRGD-TPA-DCR-NP) with fluvastatin (Fluv + NPL) for LUAD therapy. The cytotoxic effect of Fluv + NPL on LUAD cells was markedly superior to either monotherapy, demonstrating enhanced antitumor efficacy. We found that this combination not only enhanced ROS generation, mitochondrial membrane potential disruption and apoptosis, but also robustly induced ICD, as evidenced by CRT surface exposure, HMGB1 and ATP release, while concurrently inhibiting glucose uptake and glycolytic activity. Fluv + NPL suppresses DEPDC1 expression in LUAD cells, our findings identify DEPDC1 as an important regulator of glycolytic metabolism in LUAD. We demonstrate that DEPDC1 interacts with ZNF224 and is associated with NF-κB activation, which may contribute to PI3K–AKT–mTOR pathway regulation, a well-established pathway involved in glycolysis and metabolic reprogramming across diverse cancer types [19]. Beyond its role in metabolic regulation, our findings reveal that Fluv + NPL also modulates intercellular communication through DEPDC1-associated exosome-mediated signaling. As critical mediators of intercellular communication, exosomal composition changes can directly influence the phenotype and function of recipient cells [51]. Specifically, Fluv + NPL suppresses DEPDC1, which is associated with reduced exosomal CCL2 levels and altered macrophage polarization. Rescue experiments with exogenous CCL2 restored M2 macrophage polarization. Likewise, exosomes isolated from LUAD cells with CCL2 knockdown, or treatment with a CCR2 inhibitor during co-culture, similarly suppressed M2 macrophage polarization. In vivo studies further demonstrated that Fluv + NPL increased the infiltration of T cells and M1 macrophages, reduced the abundance of M2 macrophages, and improved the immune microenvironment of LUAD. Collectively, Fluv + NPL exerts a dual effect through DEPDC1 regulation by inhibiting both glycolysis and exosomal CCL2 mediated M2 macrophage polarization, thereby alleviating the immunosuppressive tumor microenvironment.

The antitumor efficacy of Fluv + NPL was validated in a subcutaneous LUAD mouse model, where it effectively improved the tumor immune microenvironment without inducing observable systemic toxicity. These findings support the therapeutic potential and biocompatibility of this strategy. However, it should be noted that subcutaneous tumor models do not fully recapitulate the anatomical architecture and immunological complexity of human lung cancer. Therefore, further investigations in more clinically relevant animal models, including orthotopic or genetically engineered LUAD models, humanized or syngeneic models with checkpoint blockade combinations, are warranted to more comprehensively evaluate anti-tumor effect and long-term immune responses. A limitation of this study is that the relatively high concentration of 200 µM fluvastatin used in vitro may limit the direct clinical translatability of our findings. Further studies are needed to determine whether optimized drug delivery strategies can achieve comparable therapeutic efficacy at lower fluvastatin doses.

Conclusion

This study shows that a high glycolysis-related risk score in LUAD is closely associated with poor prognosis and an immunosuppressive tumor microenvironment. These findings prompted us to investigate the molecular mechanisms linking glycolytic metabolism to immune regulation in LUAD. Our findings suggest that DEPDC1 is associated with glycolysis and immune modulation in LUAD. Mechanistically, DEPDC1 contributes to glycolysis through activation of the NF-κB/PI3K–AKT–mTOR signaling axis while simultaneously facilitating M2 macrophage polarization by regulating exosomal CCL2 levels. Importantly, Fluv + NPL suppresses DEPDC1 expression, resulting in inhibition of both glycolysis and exosomal CCL2-mediated macrophage polarization. Consequently, Fluv + NPL enhances T-cell and M1 macrophage infiltration, reduces M2 macrophage abundance, thereby improving the tumor immune microenvironment in LUAD. Overall, our findings identify a DEPDC1-associated glycolysis–immune regulatory axis and support the therapeutic potential of combining DEPDC1-targeted metabolic intervention with nanoparticle photodynamic therapy for LUAD treatment.