The human embryonic kidney cell line HEK293, the A549 lung adenocarcinoma cell line, and the human chronic myeloid leukemia cell line K562 were obtained from Procell Biotechnology Co., Ltd. (Wuhan, China). The A549 cell line was selected to establish the NSCLC model based on the findings of Yuan Cheng et al. (28), who demonstrated its utility in studying ICI resistance mediated by M2 macrophages. Cells were cultured in RPMI 1640 medium (Procell), which contains 10% fetal bovine serum (ExCell Bio) and 1% penicillin/streptomycin (Procell), and kept in a 37 °C incubator with 5% CO₂. All cell lines were routinely tested for mycoplasma contamination.

The cDNA sequence encoding the human monoclonal antibody for TGF-βRII (GenBank Accession: BD094925.1) was cloned into the empty vector pDC316-mCMV-EGFP (FengHui, Hunan, China). The recombinant plasmid was transformed into E. coli 5-α competent cells (New England Biolabs, NEB, USA). Positive clones were selected with ampicillin (100 μg/mL), and subsequently confirmed by Sanger sequencing (performed by FengHui). Following plasmid preparation, A549 cells were seeded in six-well plates and transfected at approximately 60% confluence using Lipofectamine 3000 (Thermo Fisher Scientific, CA, USA) according to the manufacturer’s protocol. Cells were cultured for 48 h post-transfection. Transfection efficiency was assessed by fluorescence microscopy, and anti-TGF-βRII overexpression was confirmed by western blot analysis.

The adenoviral shuttle vector GV695 comprises the following elements in sequential order: CMV-MCS-3flag-TERTp-1A(CR2del)-E1B/19K. The linearized vector was generated by restriction endonuclease digestion (NEB, USA). The target gene fragment for TGF-β inhibitor overexpression was amplified by polymerase chain reaction (PCR). Homologous recombination sequences, designed to match the ends of the linearized vector, were incorporated into the 5’ ends of the amplification primers, ensuring the PCR product’s termini were identical to the vector ends. The linearized vector and the target gene PCR product were subjected to an in vitro recombination reaction, facilitating circularization. The recombinant product was directly transformed into competent cells. Single clones were selected for PCR screening, and positive clones were subjected to sequencing and sequence analysis. Validated recombinant plasmids were expanded and purified in high concentration for use in downstream adenovirus packaging.

The genetically modified oncolytic adenovirus Ad5/3-E2F-d24 (viral backbone) was constructed as previously described (17). Expression of adenovirus E1A, essential for viral replication, is controlled by the human telomerase reverse transcriptase promoter. This design confers preferential replication in cells with elevated telomerase activity, notably cancer cells (29). A 24-base pair deletion (delta24) in the E1A gene, combined with modifications to the E2F promoter, restricts viral replication to tumor cells harboring abnormalities in the Rb/p16 pathway. The virus also carries a deletion in E1B/19K, enhancing tumor specificity. Ad-anti-TGF-βRII encodes a TGF-β inhibitor under the control of both the cytomegalovirus (CMV) promoter and the E1B promoter. Ad-null, an oncolytic adenovirus lacking transgenes, served as the control vector. HEK293 cells were co-transfected with the adenoviral shuttle plasmid (GV695) carrying the transgene and an auxiliary packaging plasmid devoid of adenoviral genomic regions (E1/E3 deletion). Recombinant, replication-selective adenovirus particles carrying the transgene were subsequently generated via Cre-loxP-mediated recombination.

Viral titers were determined using an endpoint dilution assay. HEK293 cells were seeded in a 96-well plate (1×10³ cells/well in 100 μL medium) 24 h prior to infection. The original adenovirus stock was diluted 1:100 in complete culture medium, and serial ten-fold dilutions (10^-1 to 10^-10) were then prepared. Medium was aspirated from the 96-well plate, and 90 μL of each virus dilution (10^-6 to 10^-10) was added to 10 wells per dilution. The 11th and 12th wells of each row received 90 μL of virus-free complete medium as negative controls. Plates were incubated at 37 °C with 5% CO₂ for 10 days. Cytopathic effects (CPE) were assessed, and the number of CPE-positive wells per dilution recorded. The proportion of CPE-positive wells for each dilution was calculated to determine the virus titer using the formula: Virus titer = 10^{(X + 0.8)} (PFU/mL). In this context, X represents the cumulative sum of CPE positive rates for dilutions from 10–¹ to 10^-13. This formula is applicable only when there is: (i) no CPE or growth inhibition in negative controls, and (ii) CPE in all wells at the lowest dilution.

A549 cells were seeded in 6-well plates at a density of 7×10⁵ cells/well. After 24 h, cells were infected with either the therapeutic oncolytic adenovirus (Ad-anti-TGF-βRII) or the control adenovirus (Ad-null) at a multiplicity of infection (MOI) of 50. Six hours post-infection, the medium was replaced with fresh serum-free medium, and cells were cultured for an additional 48 h. Cells were then harvested, and total RNA was extracted using RNAiso reagent (Fastagen, Shanghai, China) according to the manufacturer’s protocol. Single-stranded cDNA was synthesized from the RNA using the PrimeScript RT reagent Kit (Takara, Shiga, Japan), and gene expression was then analyzed by quantitative real-time reverse transcription polymerase chain reaction (RT-qPCR) and western blotting.

Peripheral blood mononuclear cells (PBMCs) were isolated from healthy donor blood using Lymphoprep (Solarbio, Beijing, China) density gradient centrifugation. Isolated PBMCs were washed twice with 1× phosphate-buffered saline (PBS; Servicebio Technology Co., Ltd.). PBMCs were then resuspended in Lymphocyte Serum Free Medium (88-581-CM) supplemented with 10% FBS, 100 U/mL penicillin, 10 μg/mL streptomycin, 200 IU/mL recombinant human IL-2 (PeproTech, Rocky Hill, NJ, USA), and the NK cell expansion/activation agent (LYN000101; HangZhou LuYuan Biotechnology Co., Ltd.). This mixture was then cultured in a T75 culture flask. Additional NK cell expansion/activation agent was added on day 0 and day 7. Fresh complete medium was added every 2–3 days to maintain a cell density of 0.8-1.5×10⁶ cells/mL. After 14 days of expansion, NK cells were harvested. Only NK cell populations exhibiting >90% purity (as determined by flow cytometry) and confirmed cytotoxicity were used in subsequent experiments.

Cell viability was determined using the CCK-8 assay (GLPBIO, CA, USA) according to the manufacturer’s instructions. Briefly, A549 cells were plated at a density of 1×10³ cells/well into 96-well plates and allowed to adhere overnight. Following various treatments (e.g., oncolytic adenovirus infection, co-culture with NK cells, PD-1 antibody treatment, or exposure to conditioned supernatant), 10 μL of CCK-8 solution was added per well. Plates were incubated at 37 °C for 3 h, and the absorbance at 450 nm was measured using a microplate reader (Tecan, Männedorf, Switzerland). The mean absorbance of the control group was defined as 100% viability, and values from treated groups were normalized accordingly.

Four distinct CCK-8 experiments were conducted:

After plasmid transfection or viral infection, 2.5×10⁵ cells from each group were cultured in serum-free medium for 3 h, followed by treatment with 100 ng/mL TGF-β1 (Chamot, Shanghai, China) for 30 min, with no TGF-β1 as controls. Parallelly, proteins were extracted from mouse tumor tissues. Both cellular and tissue samples were washed twice with ice-cold PBS and lysed in RIPA buffer (Solarbio, Beijing, China) supplemented with protease and phosphatase inhibitors according to the manufacturer’s specifications. Protein concentrations were quantified using a BCA assay kit (GLPBIO, CA, USA). Proteins were resolved by 10% SDS-PAGE (80 V for 30 min, then 120 V for 90 min) and electrotransferred to PVDF membranes (Immobilon^®-P, IPVH00010; MilliporeSigma, MA, USA) at 350 mA for 80 min. Membranes were blocked with 5% bovine serum albumin (BSA) in TBST (Tris-buffered saline with 0.1% Tween-20) for 2 h at room temperature, followed by overnight incubation at 4 °C with primary antibodies diluted in TBST: phospho-SMAD2 (Affinity, Cat# AF3449), SMAD2/3 (Cell Signaling Technology, Cat# 3102S), and GAPDH (Proteintech, Cat# 10494-1-AP). After three 10-minute washes in TBST, membranes were probed with horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG (Proteintech, Cat# SA00001-2) for 1 h at room temperature and washed again as described above. Protein bands were visualized using MK Hypersensitive ECL Luminescent Solution (Biomiky, Shanghai, China). Image J software was used to quantify and analyze the relative band density.

Total RNA was extracted from cell samples and murine tissues using the RNAfast200 Total RNA Extraction Kit (Fastagen Biotechnology Co., Ltd., Shanghai, China) following the manufacturer’s protocol. Extracted RNA was then reverse-transcribed into cDNA with the PrimeScript™ RT Reagent Kit (with gDNA Eraser; Takara Bio, Kusatsu, Japan), which includes genomic DNA removal prior to reverse transcription. The 20 μL reverse transcription reaction system consisted of 10.0 μL of total RNA, 1.0 μL of PrimeScript RT Enzyme Mix I, 1.0 μL of RT Primer Mix, 4.0 μL of 5× PrimeScript Buffer 2 (for Real Time), and 4.0 μL of RNase-free water. Quantitative PCR was performed using the 2×Q3 SYBR qPCR Master Mix (Universal) and the CFX Connect Real-Time System (Bio-Rad, USA). Primer sequences for RT-qPCR are listed in Table 1. GAPDH was used as an internal reference and relative expression levels were calculated using the 2^−ΔΔCT method.

NK cell culture supernatants were collected after centrifugation (500×g, 5 min). For the BALB/c nude mice tumor-bearing animal model, terminal blood was collected under isoflurane anesthesia via orbital enucleation. After 1 h incubation at room temperature, samples were centrifuged at 4 °C for 20 min. Perforin concentrations in serum and culture supernatants were measured using Human and Mouse PRF1(Perforin 1) ELISA Kit (Elabscience, Wuhan, China) according to the manufacturer’s protocol. Absorbances at 450 nm were detected with the microplate reader (Tecan, Männedorf, Switzerland).

All in vitro experiments were independently repeated at least three times unless otherwise specified. For in vivo studies, experimental group comparisons and animal cohort sizes are detailed in the corresponding figure legends. Statistical analyses were performed using GraphPad Prism v9.0 (GraphPad Software, San Diego, CA, USA). Inter-group differences were assessed by one-way ANOVA or two-way ANOVA with Tukey’s multiple comparison test. A significance level of P < 0.05 was considered significant.

A plasmid designed to overexpress a TGF-β inhibitor was constructed (Figure 1A). To validate its functionality, the plasmid was transfected into A549 cells (Supplementary Figure 2). The analysis confirmed that TGF-β signaling was reduced in transfected A549 cells, both with and without exogenous TGF-β stimulation, as shown by a significant decrease in pSMAD2 relative to total SMAD2/3, while total SMAD2/3 levels showed no significant difference (Figure 1B).

The oncolytic adenoviruses, Ad-null (control) and Ad-anti-TGF-βRII (Figure 1C), were subsequently generated after the construction of cloning vectors (Supplementary Figure 3) and viral packaging. The packaged viruses achieved high titers of 1 × 10¹¹ PFU/mL (Supplementary Figure 4).

Expanded NK cells exhibited a rapid growth phase by day 10 (Supplementary Figure 5A), culminating in a proliferation index of 350 after 14 days. Flow cytometry analysis demonstrated high purity, with >95% of cells expressing the NK cell phenotype (CD3^-CD56⁺) (Figure 1D). To assess the cytotoxic potential of the expanded NK cells, we evaluated their activity against both K562 (leukemia) and A549 (lung adenocarcinoma) cell lines. Against K562 cells in a 24 h co-culture assay, NK cells exhibited peak cytotoxic activity at an effector-to-target (E:T) ratio of 10:1, achieving approximately 60% target cell lysis (Figure 1E). For A549 cells, peak cytotoxicity also occurred at an E:T ratio of 10:1 following a 24 h co-culture (Figure 1F). (Additional morphological characterization of the expanded NK cells is provided in Supplementary Figure 5).

Both oncolytic adenoviruses, Ad-null and Ad-anti-TGF-βRII, exhibited dose-dependent cytotoxicity against A549 lung adenocarcinoma cells (Supplementary Figure 6), with Ad-anti-TGF-βRII showing significantly enhanced oncolytic activity compared to Ad-null at equivalent MOIs (Figure 1G). Notably, at an MOI of 100, Ad-anti-TGF-βRII reduced A549 cell viability to 50% at 48 h.

To investigate whether the superior cytotoxicity of Ad-anti-TGF-βRII was mediated through TGF-β pathway modulation, we analyzed anti-TGF-βRII gene expression and downstream signaling. RT-qPCR revealed a strong upregulation of anti-TGF-βRII mRNA in Ad-anti-TGF-βRII-infected cells, while minimal expression was detected in Ad-null-treated or untreated controls (Figure 1H). Consequently, western blot analysis showed a significant downregulation of the key signaling mediator pSMAD2 (Figure 1I).

These findings demonstrate that Ad-anti-TGF-βRII exerts dual anti-tumor effects: direct oncolytic activity combined with specific inhibition of the TGF-β/SMAD pathway through anti-TGF-βRII overexpression. This combined mechanism leads to enhanced tumor cell cytotoxicity and growth suppression.

When A549 cells were pre-infected with either Ad-null or Ad-anti-TGF-βRII (MOI 100, 24 h) and subsequently co-cultured with NK cells (E:T ratio 10:1, 24 h), significantly enhanced tumor cell killing was observed, with Ad-anti-TGF-βRII demonstrating superior enhancement of NK cell-mediated cytotoxicity compared to Ad-null (Figure 2A). Conditioned medium from Ad-anti-TGF-βRII-infected (MOI 100, 48 h) A549 cells also potently stimulated NK cell proliferation during a subsequent 48 h culture (Figure 2B) and altered their activation profile. These preconditioned NK cells exhibited increased intracellular granzyme B, alongside a marked reduction in intracellular perforin that corresponded with a significant rise in perforin secretion into the supernatant (Figures 2C–E), indicative of active degranulation (30, 31). Mechanistically, the intracellular granzyme B accumulation reflects both retention due to incomplete perforin pore formation during degranulation and increased de novo synthesis following NK cell activation. Additionally, they showed elevated IFN-γ mRNA levels (Figure 2F). We also examined the expression of the inhibitory receptor NKG2A and activating receptor NKp46 (32). Flow cytometry analysis demonstrated marked upregulation of NKp46 in anti-TGF-βRII-treated NK cells relative to Ad-null controls (Figures 2G, H), while NKG2A expression remained unchanged. Finally, this optimized approach demonstrated synergy with PD-1 blockade. The combination of Ad-anti-TGF-βRII pre-infection, NK cell co-culture, and PD-1 antibody treatment achieved superior tumor cell killing compared to all other conditions (Figure 2I).

Collectively, Ad-anti-TGF-βRII enhances NK cell anti-tumor activity by promoting their proliferation, upregulating NKp46, increasing granzyme B, perforin and IFN-γ production, and synergizing with PD-1 blockade.

The combination of PD-1 blockade, NK cell therapy, and oncolytic virotherapy demonstrated potent anti-tumor effects against A549 cells. While PD-1 antibody monotherapy showed limited efficacy, combination strategies significantly enhanced cytotoxic effects. Both dual-combination groups (PD-1 antibody + NK cells or PD-1 antibody + Ad-anti-TGF-βRII) demonstrated superior tumor suppression compared to single-agent treatment. Most notably, the triple-combination (PD-1 antibody + NK cells + Ad-anti-TGF-βRII) exhibited the most potent anti-tumor activity, as confirmed by microscopic examination demonstrating enhanced tumor cell killing (Figures 2J, K). We further characterized treatment-induced changes in the tumor microenvironment by analyzing key inflammatory cytokines. Measurement of IL-6 and IL-8 (33, 34) levels revealed a significant reduction in all combination therapy groups (Figures 2L, M). Although cytokine suppression in the triple-combination group was comparable to that of the PD-1 antibody + Ad-anti-TGF-βRII dual therapy, the triple-combination approach was still considered as the most effective anti-tumor strategy in vitro due to its superior direct tumoricidal activity.

The combined therapeutic strategy of PD-1 blockade, NK cell immunotherapy and TGF-βRII-targeted oncolytic virotherapy demonstrated significant antitumor efficacy in A549 xenograft models. When treatments began on tumors of approximately 100 mm³ (Figure 3A), the PD-1 antibody monotherapy only inhibited tumor growth by 28.83%. In contrast, the combination therapies were substantially more effective: adding NK cells achieved 74.98% inhibition, while adding Ad-anti-TGF-βRII achieved 77.32% inhibition. Notably, the triple combination therapy (PD-1 antibody + NK cells + Ad-anti-TGF-βRII) exhibited superior tumor suppression with 93.09% growth inhibition (P < 0.01 versus all other groups; Figures 3B–E), corresponding to a 79.78% reduction in final tumor weight compared to controls (187.6 mg versus 927.6 mg; Figure 3F). All treatment regimens were well-tolerated, with no significant body weight changes observed (Figure 3G). Histopathological analysis confirmed that the triple combination induced pronounced inflammatory infiltration and extensive tumor necrosis (Figure 3H). These findings highlight the synergistic potential of combining immune checkpoint inhibition with cellular immunotherapy and targeted virotherapy for NSCLC treatment.

Analysis of tumor tissues confirmed the in vivo activity of the therapeutic transgene and its target. Anti-TGF-βRII mRNA was detected exclusively in virus-treated groups, and correspondingly, TGF-β/SMAD signaling (pSMAD2) was most strongly suppressed in the triple-combination group (Figures 4A, B). Collectively, these data confirm successful in vivo transgene delivery, significant suppression of pSMAD2 in treated tumors, and enhanced TGF-β pathway blockade with combination therapy, aligning with our in vitro observations.

The triple-combination therapy demonstrated superior immunostimulatory effects compared to all other treatment groups. Serum analysis revealed significantly elevated levels of cytotoxic mediators (perforin:10376.9 ± 346.9 pg/mL; GZMB: 7244.6 ± 747.9 pg/mL and IFN-γ: 119.6 ± 6.6 pg/mL) (all P < 0.01 vs monotherapy; Figure 4C). Corresponding increases were observed in tumor microenvironment cytokines, with sFas L (4063.8 ± 1124.3 pg/mL) and IL-2 (359.0 ± 39.0 pg/mL) showing the highest concentrations in the triple therapy group (Figure 4D). Transcriptional profiling of tumor tissues revealed marked upregulation of perforin (28.3 ± 1.2-fold, P < 0.0001), GZMB (32.1 ± 5.7-fold, P < 0.001), and IFN-γ (89.3 ± 20.9-fold, P < 0.01), alongside a reduction in HIF-1α (47.6% ± 8.3%, P < 0.05) (Figure 4E). These molecular changes indicate enhanced antitumor immune responses coupled with improved tumor microenvironment conditions.

Analysis of immune cell infiltration revealed distinct patterns. Immunofluorescence (IF) staining for CD3^-CD56⁺ cells confirmed the specific recruitment of NK cells into tumor tissues receiving combination therapies, with the most intense signal in the triple-combination group (Figures 5A, B). Similarly, immunohistochemistry (IHC) showed that infiltration of CD4⁺ and CD8⁺ T cells also peaked in this group (Figures 5C, D), while PD-1 expression was substantially diminished (Figures 5E, F).

These results demonstrate that the triple combination therapy orchestrates a coordinated immune response through enhanced NK cell recruitment and cytotoxicity, promoted effector T cell infiltration, and reduced PD-1-mediated immunosuppression.

To evaluate the safety profile of the combination immunotherapy (PD-1 antibody + NK cells + Ad-anti-TGF-βRII), we first assessed the tissue specificity of the oncolytic adenovirus. Using the virally encoded anti-TGF-βRII transgene as a molecular tracer, RT-qPCR analysis demonstrated restricted viral replication to tumor tissues, with no detectable anti-TGF-βRII mRNA in normal organs including liver, lungs, or heart (Figure 6A). These results confirm the tumor-selective tropism of Ad-anti-TGF-βRII and its inability to disseminate systemically.

Histopathological analysis of major organs (liver, heart, kidneys, and lungs) showed preserved tissue architecture with no evidence of treatment-related abnormalities following the combination therapy (Figure 6B). The absence of pathological changes in these critical organs indicates a favorable safety profile for this multimodal regimen, which includes intratumoral viral administration, intravenous NK cell delivery, and systemic PD-1 antibody treatment. Correspondingly, serum levels of injury markers (LDH and ALT) were significantly reduced in the treatment group (Figures 6C, D). These findings, together with the normal hepatic architecture observed in H&E-stained sections (Figure 6B), demonstrate the absence of treatment-induced hepatotoxicity.

To synthesize these findings into a unified model, we propose a mechanistic framework for the observed synergy (Figure 6E). Our data demonstrate that, in addition to its direct oncolytic effect, intratumoral administration of Ad-anti-TGF-βRII remodels the TME by blocking TGF-β signaling, which in turn primes adoptively transferred NK cells and enhances their cytotoxic function. This effect is further amplified by systemic PD-1 blockade, which synergistically overcomes ICI resistance while maintaining a favorable safety and efficacy profile.