The human NSCLC A549 (RRID:CVCL_0023), H1299 (RRID:CVCL_0060), H460 (RRID:CVCL_0459), H1975 (RRID:CVCL_1511) and H2170 (RRID:CVCL_1575) cell lines, along with the normal human bronchial epithelial cell line (BEAS-2B; RRID:CVCL_0168), were from Procell Life Science & Technology Co., Ltd. and maintained in appropriate medium supplemented with 10% fetal bovine serum (FBS; Wuhan Pricella Life Science and Technology Co., Ltd.) and 50 μg/ml penicillin-streptomycin (Gibco; Thermo Fisher Scientific, Inc.) in the appropriate conditions. Recombinant human TGF-β1 was from PeproTech, Inc. and stored as recommended.

Gene Expression Profiling Interactive Analysis (GEPIA; http://gepia.cancer-pku.cn) was used to predict the levels of gene expression in NSCLC and normal tissues. It was also employed to assess the correlation between genes. Protein-protein interaction (PPI) networks were constructed using the STRING database (http://string.embl.de/). Using the Kaplan-Meier plotter (http://kmplot.com/analysis/), the relationship between Notch3 expression and survival in lung cancer was examined. Additionally, the log-rank P-value and the hazard ratio (HR) with 95% confidence intervals were calculated.

Proteins were extracted from cells with RIPA buffer (Beijing Solarbio Science & Technology Co., Ltd.) with 1 mM PMSF protease inhibitor (MedChemExpress) added. A BCA Protein Assay Kit (Beyotime Institute of Biotechnology) was used to determine the concentration of total proteins in accordance with the manufacturer's instructions. A total of 40 μg protein was separated by 10% SDS-PAGE and then transferred to polyvinylidene difluoride membranes (MilliporeSigma). After the membranes were blocked with 5% milk (Beyotime Institute of Biotechnology) in tris-buffered saline containing 0.1% Tween-20 (TBST) at room temperature for 1 h and then incubated at 4°C overnight with primary antibodies, including anti-Notch3-antibody (1:2,000; cat. no. CL488-55114; Proteintech Group, Inc.), anti-GAPDH antibody (1:10,000; cat. no. 60004-1-Ig; Proteintech Group, Inc.), anti-Vimentin (VIM) antibody (1:10,000; cat. no. 10366-1-AP; Proteintech Group, Inc.), anti-E-Cadherin antibody (1:5,000; cat. no. 31863-1-AP; Proteintech Group, Inc.), anti-β-actin antibody (1:2,000; cat. no. 60008-1-Ig; Proteintech Group, Inc.), anti-ALDH1A1 antibody (1:10,000; cat. no. 83830-2-RR; Proteintech Group, Inc.), anti-OCT4 antibody (1:2,000; cat. no. 60242-1-Ig; Proteintech Group, Inc.), anti-CD44 antibody (1:2,000; cat. no. 60224-1-Ig; Proteintech Group, Inc.) and anti-CD133 antibody (1:2,000; cat. no. 18470-1-AP; Proteintech Group, Inc.). The next day, the protein was washed with TBS-0.1% Tween three times and then incubated with secondary antibodies conjugated with horseradish peroxidase conjugated goat anti-rabbit IgG (1:10,000; cat. no. SA00001-2; Proteintech Group, Inc.) or anti-mouse IgG (1:10,000; cat. no. SA00001-1; Proteintech Group, Inc.) for 1 h at room temperature and then visualized by ECL detection kit (cat. no. BMU102-CN; Abbkine Scientific Co., Ltd.). Protein visualization was performed using ECL HRP substrate (cat. no. 34580; Thermo Fisher Scientific, Inc.). Images were captured using the BIO-RAD ChemiDoc MP multifunctional chemicalimaging instrument (Bio-Rad Laboratories, Inc.). The expression of protein was analyzed by ImageJ software (ImageJ 2.0; National Institutes of Health).

Total RNA was extracted from the cells using the FreeZol reagents (cat. no. R711-01; Vazyme Biotech Co., Ltd.) according to the manufacturer's instructions. The RNA concentration was measured using a Nano Drop ONE spectrophotometer (Thermo Fisher Scientific, Inc.). cDNA was performed according to the manufacturer's protocol using 1 μg total RNA and PrimerScript RT reagent kit (cat. no. R222-01; Vazyme Biotech Co., Ltd.). RT-qPCR was performed with SYBR Green Master Mix (cat. no. Q412-02; Vazyme Biotech Co., Ltd.). The thermocycling conditions were as follows: Initial denaturation at 95°C for 30 sec. Denaturation at 95°C for 10 sec, annealing and extension at 60°C for 30 sec, 40 cycles. All reactions were conducted in a 20 μl reaction volume in triplicate. The primers used in these experiments are listed in Table I. The relative level of RNA was normalized to that of GAPDH using the 2^−ΔΔCq method (23).

The 3rd generation system was used to generate lentivirus. Recombinant packaging plasmids (PG-P1-VSVG, PG-P2-REV and PG-P3-RRE) and vector plasmids (lentivirus vector) were prepared by Shanghai GenePharma Co., Ltd. A total of 1 μg third generation lentiviral packaging system package mix was prepared in the ratio of PG-P1-VSVG: PG-P2-REV: PG-P3-RRE 1:2:3 for the lentiviral plasmid packaging experiments in 60-mm cell culture dishes. RNAi-Mate (cat. no. GO4001; Shanghai GenePharma Co., Ltd.) to co-transfect 293T cells (Shanghai GenePharma Co., Ltd.). Cells were incubated at 37°C with 5% CO₂ for 48 h post-transfection. After 48 h, the viral supernatant was collected, filtered through a 0.45-μm filter and concentrated by ultracentrifugation at 73,000 × g for 2 h at 4°C. ShNotch3-containing plasmid backbone was LV3(H1/GFP&Puro) and its sequence was 5′-AGATCTGGCACCCTGACTTAT-3′. Sequence of shSUCLG1 negative control was 5′-TTCTCCGAACGTGTCACGT-3′. Short hairpin (sh)RNA sequences used were shRNA sequence targeting Notch3: Sh-Notch3-1:5′-GGGTTTGAGGGTCAGAATTGT-3′, Sh-Notch3-2:5′-GATGCTATCTGTGACACAAAT-3′, Sh-Notch3-3:5′-GCTTGGGAAATCAGCCTTACA-3′; and a scrambled shRNA sequence (5′-TTCTCCGAACGTGTCACGT-3′) was used as control. A549 and H1975 cells (5×10⁵ cells/well) were seeded in six-well plates and cultured for 24 h at 37°C to reach 80% confluence. Lentiviral particles were mixed with polybrene (8 μg/ml) at a multiplicity of infection of 50, then added to cells and incubated at 37°C for 48 h. Puromycin (1 μg/ml) was added 72 h post-transfection to facilitate screening of stably infected cells and 2-3 days later for subsequent experiments (1 μg/ml puromycin used for maintenance). For Notch3 overexpression, the pCLE Notch3 intracellular domain (N3ICD) over (pCLE NICD3, RRID: Addgene-26894) was purchased from Addgene, Inc. Cells were transfected with 2.5 μg NICD3 plasmid per well (six-well plate) using Lipofectamine^® 3000 with Opti-MEM (cat. no. 31985-062; Invitrogen; Thermo Fisher Scientific, Inc.) in a biological safety cabinet, based on the kit's instructions. After 6 h of incubation at 37°C, the medium was replaced with fresh complete culture medium. A total of 48 h after infection, the transfection efficiency was confirmed by RT-qPCR and WB.

The levels of TGF-β1 in the culture were measured using an ELISA kit (cat. no. EK0513; Boster Biological Technology, Ltd.), following the manufacturer's guidelines.

Lung cancer cells were cultured in 96-well plates (2,000 cells/well). According to the manufacturer's instructions, 10 μl of CCK-8 solution was added to each well and incubated at 37°C for 3 h. Absorbance at 450 nm was measured.

A549 and H1975 cells were seeded into 6-well plates. When cells reached 90-95% confluence, the monolayer was scraped with a sterile 100 μl pipette tip and dislodged cells were removed with PBS. The cells were cultured in serum-free medium at 37°C for 48 h, and images were acquired using a Zeiss light microscope (Carl Zeiss AG). The wound area at the same location was subsequently measured by ImageJ (V1.54d, National Institutes of Health). The cell migration rate was calculated as follows: Cell migration rate (%)=(initial wound area-wound area after 48 h)/initial wound area ×100%.

Cells were plated in 24-well Transwell chambers (Corning, Inc.), with the lower chambers containing 600 μl of complete medium. After 48 h, the cells in the upper chamber were removed using a cotton swab. The migrated or invaded cells were fixed with 4% paraformaldehyde at room temperature for 15 min, stained with 0.1% crystal violet for 30 min at room temperature and quantified manually using a Zeiss light microscope (Carl Zeiss AG).

The RNA sequencing service was provided by CloudSeq Biotech Inc., which extracted RNA from cells using FreeZol reagent (Vazyme Biotech Co., Ltd.), with its quality and concentration evaluated using a NanoDrop-1000 spectrophotometer (Thermo Fisher Scientific, Inc.). rRNA was removed using an GenSeq^® rRNA Removal Kit (GenSeq, Inc.) according to the manufacturer's instructions. RNA samples after rRNA removal were passed through the GenSeq^® Low Input RNA Library Prep Kit (GenSeq, Inc.) to construct an RNA sequencing library. A BioAnalyzer 2100 instrument (Agilent Technologies, Inc.) was used for library quality control and quantification. 150 bp double-ended sequencing was performed on an Illumina NovaSeq instrument (Illumina, Inc.). Q30 (threshold for high-quality bases, with error rate <0.1%) was used for quality control. Cutadapt software (24) (v1.9.3) was used to remove splices, subsequently low-quality reads were removed and high-quality reads were obtained. The filtered clean reads were then aligned to the reference genome with HISAT2 software (25) (v2.0.4). HTSeq software (26) (v0.9.1) was then used to generate the raw count and the edgeR (27) package was used to perform normalization and examine differentially expressed genes with a |log2 (fold change)| of ≥1 and false discovery rate of ≤0.05. Gene Ontology (GO) and pathway enrichment analysis were performed using the clusterProfiler package (27,28). Sequencing data have been uploaded to the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) (https://www.ncbi.nlm.nih.gov//bioproject/PRJNA1275425) under accession number (PRJNA1275425).

A549 cells were harvested and counted and 5,000 cells in serum-free DMEM-F12 (Gibco; Thermo Fisher Scientific, Inc.) containing EGF (20 ng/ml; PeproTech, Inc.), b-FGF (10 ng/ml; PeproTech, Inc.) and B27 (1:50 dilution, BD Biosciences) were plated in each well of six-well ultralow attachment plates (Corning, Inc.). After one week, the former tumor spheres were counted under a phase-contrast microscope (Carl Zeiss AG).

A549 cells (1×10⁴ cells/well) were grown on coverslips in 24-well plates at 37°C for 24 h. Cells were fixed by 4% paraformaldehyde for 30 min, infiltrated with 0.2% Triton X-100 for 10 min, blocking with 5% bovine serum albumin (Beyotime Institute of Biotechnology) for 1 h at room temperature. Diluted CD133 antibodies (1:200 cat. no. 18470-1-AP; Proteintech Group, Inc.), ALDHA1 (1:200; cat. no. 83830-2-RR; Proteintech Group, Inc.), Vimentin (1:200; cat. no. 10366-1-AP; Proteintech Group, Inc.) and E-Cadherin (1:200; cat. no. 31863-1-AP; Proteintech Group, Inc.) antibodies were placed on glass slides and incubated overnight at 4°C. Next, the coverslips were treated to Alexa Fluor 488 goat anti-rabbit IgG (1:200; cat. no. BA1127; Boster Biological Technology, Ltd.) or Alexa Fluor 594 goat anti-rabbit IgG (1:200; cat. no. BA1142; Boster Biological Technology, Ltd.) at room temperature for 1 h in the dark. Cells were washed and stained with 4′,6-diamidino-2-phenylindole (DAPI, 1 mg/ml; 1:5,000; cat. no. S2110; Beijing Solarbio Science & Technology Co., Ltd.) for 10 min at room temperature. The images were imaged under a fluorescent microscope (Zeiss Axio Observer Z1 LSM710; Carl Zeiss AG).

A total of 44 nude mice (female; aged 4-6 weeks; weight, 10-14 g) were acquired from the Comparative Medicine Center of Yangzhou University (Yangzhou, China; approval no.202406015) and raised in a specialized pathogen-free environment (21-26°C; humidity, 40-70%; 12-h light/dark cycle and ad libitum access to food and water).

The mice were randomly divided into eight groups (4 groups for the xenograft models and 4 groups for the metastasis models). A suspension of 5×10⁶ A549 cells in 200 μl of PBS-Matrigel (1:1 ratio) was subcutaneously injected in the right underarm of mice. Mice were randomly divided (six mice/group) into four groups as follows: i) sh-NC group; ii) sh-NC+TGF-β1 group (sh-NC A549 cells pre-treated with 10 ng/ml TGF-β1 for 48 h before injection); iii) sh-Notch3 group; iv) sh-Notch3+TGF-β1 group (sh-Notch3 A549 cells pretreated with 10 ng/ml TGF-β1 for 48 h before injection). Tumor size was measured once a week with calipers and tumor volumes were calculated by the following formula: Volume (mm³)=0.5 × Length (mm) × Width² (mm²). Animals were checked daily and any animal found unexpectedly to be moribund, cachectic or unable to obtain food or water was sacrificed. In strict accordance with animal ethical regulations, the weight of tumors did not exceed 10% of the mouse body weight, the average tumor diameter did not exceed 20 mm, and the tumor volume did not exceed 2,000 mm³. After 5 weeks, all animals were euthanized using a 30-70% per min displacement of chamber air with compressed CO₂. After the completion of euthanasia with CO₂ inhalation, cervical dislocation with the confirmation of a gap between the skull and spinal column was used to verify the death of mice. All tumors were dissected from mice and the total tumor weight in each mouse was measured. Some tumors were fixed in 4% formalin for 24 h at room temperature for subsequent immunohistochemistry. Additionally, the tumor metastasis model was created by injecting A549 cells into the tail veins of additional BALB/c nude mice. To perform this, 1×10⁶ cells mixed with PBS were injected into each mouse in 4 groups (n=5 per group). No metastatic nodules were found in the abdominal and thoracic organs of the subcutaneous xenograft mice. After 2 months, the mice were euthanized with excessive carbon dioxide and the lungs of the nude mice were extracted to observe whether there were metastatic lesions and stained with hematoxylin and eosin (H&E). After euthanasia, the death of the animals was confirmed based on the disappearance of pain response, no response when pressing the toes with hands or forceps and observation of cardiac and respiratory arrest. No nude mice experienced unexpected mortality during the experiment. All mice were euthanized at the planned end of the experiment.

The tumor tissues collected from a mouse xenograft model were fixed in 4% paraformaldehyde at 4°C for 24 h and then embedded in paraffin. Sections of paraffin-embedded tissue (2-3 μm) were dried for 60 min at 60°C and then subjected to two rounds of dewaxing in xylene for 30 min each. Subsequently, a decreasing gradient of alcohol concentrations was used for hydration. For antigen retrieval, the sections were heated in citrate buffer at 100°C for 10 min. The sections were blocked with 5% BSA for 1 h at room temperature, and 3% H₂O₂ was used to block endogenous peroxidase/phosphatase activity for 10 min at room temperature. Subsequently, the sections were incubated overnight at 4°C with the following primary antibodies: Anti-Vimentin (VIM) (1:5,000; cat. no. 10366-1-AP), anti-E-Cadherin antibody (1:500; cat. no. 31863-1-AP.), anti-CD44 antibody (1:500; cat. no. no. 60242-1-Ig) and anti-CD133 antibody (1:5,000; cat. no. 18470-1-AP) and anti-Ki67 (1:5,000; cat. no.27309-1-AP; all from Proteintech Group, Inc.). The sections were then treated with secondary antibodies labeled with biotin (1:1,000; cat. no. SA00001-2; Proteintech Group, Inc.) for 1 h at room temperature and incubated with an avidin solution labeled with horseradish peroxidase (HRP; Proteintech Group, Inc.) for 10 min at room temperature. The DAB staining solution was incubated at room temperature for 15 min, and then hematoxylin was applied as the counterstain at room temperature for 3 min. Finally, the sections were dried by progressively adding ethanol, as well as xylene, and were subsequently sealed using neutral balsam. The results were observed using a light microscope.

The tissues were deparaffinized with xylene and immersed in an EDTA antigen retrieval buffer. Then, the tissues were stained with H&E. The sections were stained with hematoxylin dye for 2 min at room temperature, treated with 1% hydrochloric acid alcohol for 45 sec at room temperature, stained with eosin dye for 2 min at room temperature, dehydrated with ethanol (75, 85, 95 and 100%), washed with xylene and finally sealed with neutral resin. The results were observed under a light microscope.

The statistical analyses were performed using GraphPad 7.0 (GraphPad Software, Inc.; Dotmatics). Comparisons between two groups were performed using unpaired Student's t-test, while multi-group comparisons were analyzed using one-way ANOVA followed by Tukey's multiple comparison test. Data were expressed as means ± SD. P<0.05 was considered to indicate a statistically significant difference.

The present study first examined the expression of Notch3 in NSCLCs and bronchial epithelial cells (BEAS-2B). Western blotting and RT-qPCR showed that Notch3 protein and mRNA levels were highly expressed in the non-small cell lines and BEAS-2B (Fig. 1A and B). To evaluate the expression level of Notch3 in pathological tissues, the present study performed an expression analysis by comparing lung adenocarcinoma (LUAD) and lung squamous cell carcinoma (LUSC) to normal tissues by using the GEPIA server. It was found that Notch3 was clearly increased in NSCLC tissues (Fig. 1C). Meanwhile, the human protein atlas verified an obvious expression trend of Notch3 protein within NSCLC tissues (Fig. 1D).

Next, the relationship between Notch3 expression and prognosis in lung cancer patients was evaluated using the Kaplan-Meier plotter database. The results indicated that higher Notch3 expression was markedly associated with a poorer prognosis in lung cancer, with overall survival (OS) HR=1.3 [95% confidence interval (CI)=1.16-1.47; P=1.4×10⁻⁵] and first progression (FP) HR=1.32 (95% CI=1.12-1.57; P=1.1×10⁻³). However, Notch3 expression did not show a significant correlation with the prognosis of lung cancer patients in terms of post-progression survival (PPS; PPS HR=1.07, 95% CI=0.87-1.31, P=0.54) (Fig. 1E).

Notch3 intracellular domain (N3ICD) overexpression plasmid was transfected into A549 and H1975 cells and its expression was confirmed by WB. Compared with the empty vector-transfected cells, the protein levels of N3ICD were markedly increased in the plasmid-transfected cells (Fig. 2A). Lentivirus was employed to establish stable cell lines with downregulated (shNotch3) expression levels of Notch3 in A549 cells. Among the three virus sequences, the silencing efficiency of Notch3 was assessed using RT-qPCR and WB (Fig. 2C and D) and selected the sequence with the highest silencing efficiency for further experiments. When N3ICD was upregulated in lung cancer cells, CCK-8 assay results showed that the viability of lung cells was enhanced, whereas Notch3 knockdown resulted in decreased cell viability (Fig. 2B and E).

The present study then compared the transcriptomes between A549 cells with control (shNC) and Notch3 knockdown (shNotch3) by RNA sequencing. KEGG analysis revealed that the TGF-β signaling pathway was markedly affected by Notch3 knockdown (Fig. 3A). Volcano plots, generated using P-values and fold changes, helped visualize the differential gene expression between the control and sh-Notch3 groups. The RNA sequencing results showed that 345 genes were upregulated, while 321 were downregulated (Fig. 3B). Additionally, hierarchical clustering revealed distinct differences between the control and sh-Notch3 groups (Fig. 3C). Moreover, by constructing PPI networks, associations between Notch3 and TGF-β1 were identified, as well as with EMT-associated proteins such as VIM and CDH1 (Fig. 3D). The GEPIA server was used to analyze the correlation between Notch3 and TGF-β1 in normal lung tissues and lung cancer, as well as EMT-related markers. TGF-β1 and VIM were markedly positively associated with Notch3, while E-Cadherin (CDH1) showed a negative correlation (Fig. 3E). These findings suggested a strong association between Notch3 and EMT markers.

Given that TGF-β1-induced EMT is well-established in cell lines such as H1975, H1299 and A549, the role of Notch3 in this process was further investigated. To confirm this, overexpressed N3ICD plasmids and Notch3-knockdown lentiviruses were transfected into A549 and H1975 cells. The results showed that TGF-β1 secretion increased upon N3ICD overexpression, while TGF-β1 secretion decreased after Notch3 knockdown (Fig. 3F). This suggests that Notch3 may promote EMT by upregulating TGF-β production. Western blotting further confirmed the relationship between Notch3 and EMT markers in both cell types. After transfecting the N3ICD overexpression plasmid and Notch3-silencing siRNA into A549 and H1975 cells and treating them with TGF-β1 (Fig. 3G), N3ICD overexpression enhanced TGF-β1-induced changes in EMT marker expression, resulting in a marked increase in VIM levels and a reduction in E-Cadherin expression. By contrast, silencing Notch3 reversed these TGF-β1-mediated changes in EMT marker expression.

The present study investigated the role of Notch3 in TGF-β1-induced EMT in lung cancer by silencing or overexpressing N3ICD in lung cancer cells. The results showed that increased Notch3 expression in A549 and H1975 cells markedly enhanced cell proliferation, while reduced Notch3 levels inhibited cell proliferation. Notably, TGF-β1 itself did not have a significant effect on cell proliferation (Fig. 4A). Transwell and wound healing assays revealed that Notch3 knockdown reversed the inhibitory effect of TGF-β1 on cell migration and invasion, whereas N3ICD overexpression had the opposite effect, promoting cell migration and invasion in lung cancer cells (Fig. 4B and C). In conclusion, these results suggest that Notch3 promotes EMT by activating TGF-β1.

One of the effective acquisition methods for CSCs is SFM suspension culture. Based on the experimental requirements, the human lung cell line A549 was cultured in SFM culture conditions to enrich related CSCs. After three passages in culture, numerous spheroid clusters were observed using microscopy (Fig. 5A). CD133 and ALDH1A are two of the most well characterized biomarkers of lung cancer (29). Immunofluorescence staining showed that sphere cells exhibited a higher expression of CD133 and ALDH1A Compared with parental cells (Fig. 5B). To further investigate the difference in cancer stemness between parental and sphere cells, the cells were collected and the expression of stem cell transcription factors and stem cell markers assessed. Sphere cells showed markedly higher expression levels of Oct4, Sox4, CD133 and Nanog Compared with parental cells (Fig. 5C). Additionally, WB of stem cell markers confirmed that sphere cells exhibited elevated levels of CD133, CD44, OCT4 and ALDH1A Compared with parental cells (Fig. 5D). Enhanced tumor invasion, a hallmark of CSCs, was observed in the Transwell assay (Fig. 5E). Taken together, these results indicated that the sphere cells derived from A549 cells were successfully enriched and exhibited CSC characteristics, making them suitable for further experimentation.

Activation of the EMT is an important process for the transformation of common tumor cells into CSCs, which in turn promotes the formation of CD133+/ALDH1+ subpopulations. EMT is also closely linked to tumor progression, metastasis and drug resistance. It was observed that the expression of the epithelial marker E-Cadherin was markedly downregulated in A549 spheres Compared with parental cells, while the mesenchymal marker VIM was markedly upregulated (Fig. 6A-C). Furthermore, TGF-β1 secretion was notably increased in A549 spheres Compared with parental cells (Fig. 6D). These findings suggested that the formation of cancer stemness may be accompanied by the activation of EMT. The expression of Notch3 was next examined using WB. The results showed that Notch3 expression was elevated in A549 spheres (Fig. 6E). Based on these observations, it was concluded that Notch3 is associated with A549-derived CSCs.

The present study investigated the role of Notch3 in regulating self-renewal by assessing the sphere formation ability. It was found that Notch3 knockdown led to a decrease in spheroid formation in A549 cells after TGF-β1 stimulation, Compared with control cells. By contrast, N3ICD overexpression resulted in an increased number of spheroids following TGF-β1 stimulation (Fig. 7A). Additionally, WB showed that overexpression of Notch3 markedly enhanced the TGF-β1-induced upregulation of stemness markers. However, inhibition of Notch3 expression markedly reduced the TGF-β1-induced increase in stemness marker expression (Fig. 7B). These results strongly suggest that Notch3 positively regulates the stemness of lung cancer cells through TGF-β1.

To confirm whether Notch3 expression could inhibit the tumorigenesis of lung cells in vivo, A549 cells infected with a specific shRNA targeting Notch3 or a mock shRNA control were implanted into BALB/c nude mice and TGF-β1 was injected intraperitoneally as described in the Materials and methods section. The tumor volume was examined every 7 days. Compared with the control group, the TGF-β1 group demonstrated an increase in tumor volume, whereas a decreased tumor volume was observed in the Sh-Notch3 group. After 35 days, the xenografts were exercised from the mice and similar changes in tumor volume and weight were observed. These findings suggested that Notch3 antagonized TGF-β1-induced tumor growth (Fig. 8A).

The protein expression levels of E-Cadherin, VIM, CD133, CD44 and Ki-67 in the transplanted tumor tissues were assessed using immunohistochemistry. Sh-Notch3 reduced the protein levels of VIM, Ki-67, CD44 and CD133 and increased E-Cadherin expression. By contrast, the TGF-β1 group exhibited the opposite effects. Notably, sh-Notch3 reversed the inhibitory effect of TGF-β1 on E-Cadherin expression and attenuated the TGF-β1-induced upregulation of VIM, CD44, CD133 and Ki-67 (Fig. 8B). Next, a metastatic model was established by injecting A549 cells via the tail vein of the mice, followed by intraperitoneal TGF-β1 injection. Compared with the negative control group, the Notch3 knockdown group exhibited a decrease in the number of lung metastatic nodules, while TGF-β1 induction markedly increased the number of metastatic nodules. Interestingly, Notch3 knockdown decreased the number of TGF-β1-induced lung metastatic nodules (Fig. 8C). These results confirm that Notch3 knockdown inhibits TGF-β1-induced tumorigenesis and metastasis of A549 cells in vivo.