A549 and SK-MES-1 cell lines were purchased from the Chinese Academy of Medical Sciences, Shanghai, China. A549 cells were cultured in F12K medium (Gibco/Invitrogen, Gaithersburg, MD), and SK-MES-1 cells were cultured in MEM medium (Gibco/Invitrogen). All media were supplemented with 10% fetal bovine serum (FBS; Gibco/Invitrogen), 100 units/mL penicillin, and 100 mg/mL streptomycin (Gibco BRL). Cells were cultured at 37°C in a humidified incubator equilibrated with 5% CO₂. Transfection was performed using Lipofectamine RNAiMAX reagent (Invitrogen, Carlsbad, CA) according to the manufacturer’s instructions. D-Luciferin Firefly was purchased from Perkin Elmer (Waltham, MA). PI3K inhibitor was purchased from Beyotime (Nantong, China). CRISPRa kit was purchased from OriGene (cat No. GA105405). The siRNAs were purchased from Sigma-Aldrich (St. Louis, MO). siRNA sequences are listed in S1 Table. PI3K inhibitor (ZSTK474), epidermal growth factor receptor (EGFR) inhibitor (gefitinib, osimertinib, erlotinib) was purchased from MCE MedChemExpress. All the experiments were performed in triplicate and repeated at least three times.

Recombinant lentiviruses were constructed by Shanghai GenePharma (Shanghai, China). Concentrated viruses were used to infect 5×10⁵ cells in a 60-mm dish with 8 μg/mL polybrene. Infected cells underwent sorting for target expression. shRNA sequences are listed in S1 Table.

The antibodies used were anti-MLL4 (Sinalway Antibody, Baltimore, MD), anti-AKT, anti–p-AKT (Cell Signaling Technology, Danvers, MA), anti-Histone H3, anti-SOX2 (Abcam, Cambridge, MA), anti-MLL4, anti-PIK3IP1, anti-THBS1, anti-β-actin (Proteintech, Rosemont, IL), anti-MLL4 (Sigma-Aldrich), and anti-p53 (Santa Cruz Biotechnology, Santa Cruz, CA).

Total RNA was extracted and determined by real-time quantitative reverse transcription polymerase chain reaction (RT-qPCR) as previously reported [18]. β-Actin was set as the internal control. The primers used are listed in S2 Table.

Cellular extracts were harvested and incubated with the appropriate primary antibody or normal mouse/rabbit IgG at 4°C overnight. Samples were mixed with protein A/G magnetic beads for 2 hours at 4°C, and following a wash, the beads underwent sodium dodecyl sulfate polyacrylamide gel electrophoresis, followed by immunoblotting with a secondary antibody. Immunodetection was performed using enhanced chemiluminescence with an enhanced chemiluminesence system (cat. No. WBKLS0500, Millipore, Billerica, MA) according to the manufacturer’s instructions.

Chromatin immunoprecipitation (ChIP) assay was performed in A549 cells as described previously [19]. The primer sequences are listed in S3 Table.

Briefly, cells were cultured with 5-ethynyl-2′-deoxyuridine (EdU) for 2 hours before detection. The proliferation rate of the cells was then evaluated using a Cell-Light EdU Cell Proliferation Detection kit (RiboBio, Guangzhou, China) following the manufacturer’s instructions.

Briefly, 2×10⁴ cells were seeded into the upper chamber in a volume of 300 μL. Cells on the upper side of the membrane were removed by cotton swabs and those on the other side were stained and counted.

Briefly, cells were grown to confluence and wounds were made using sterile pipette tips. Cells were washed with phosphate buffered saline (PBS) and incubated in the fresh medium without FBS. After 36 hours of incubation at 37°C, the cells were imaged. Data shown are the mean±standard deviation (SD) for n=6 wells per group.

CRISPR activation (CRISPRa) is one type of CRISPR tool that uses the enzymatically deficient Cas9, a mutation of Cas9 without endonuclease activity, CRISPRa system can be used to activate endogenous gene expression. Approximately 18–24 hours before transfection, 3×10⁵ adherent cells were plated in 2 mL culture media into each well of a 6-well plate. One microgram of MLL4 gRNA vector (or Scramble gRNA) and 0.3 μg enhancer vector were diluted in 250 μL of Opti-MEM and gently vortexed. And 5 μL of Turbofectin was added to the diluted DNA (not the reverse order) and pipetted gently to mix completely. The mixture was incubated for 15 minutes at room temperature. Then, the mixture was added drop-wise onto the plated cells. Forty-eight hours post-transfection, gene expression can be measured with quantitative real-time PCR (qPCR).

A549 or SK-MES-1 cells were inoculated into a 1:1 mixture of DMEM/F12 medium (Gibco/Invitrogen) without serum containing B27 supplement (50×, Invitrogen), 0.4% BSA, 4 mg/mL insulin (Invitrogen), human recombinant epidermal growth factor (20 ng/mL), and basic fibroblast growth factor (10 ng/mL) at a density of 1,000 cells/well in ultra-low attachment 6-well plates (Corning, Inc., Corning, NY). The size and number of spheroids were analyzed under an optical microscope. Spheroids more than 30 μm in size are the criteria for sphere formation.

All human tissue samples were collected using protocols approved by the Ethics Committee of Lishui Hospital, and informed consent was obtained from all patients. Samples were frozen in liquid nitrogen immediately after surgical removal and maintained at −80°C until analysis. Samples were fixed in 4% paraformaldehyde (Sigma-Aldrich) at 4°C overnight, embedded in paraffin, sectioned at 8 μm onto Superfrost-Plus Slides, which were then processed per standard protocols using 3,3’-diaminobenzidine staining and monitored microscopically.

4×10⁶ A549 cells in 100 μL of PBS were injected subcutaneously into the 6–8-week-old BALB/c nude mice (Vital River, Beijing, China). Five animals per group were used in each experiment. Tumors were measured every 3 days using a vernier caliper, and the volume was calculated according to the following formula: π/6×length×width². All studies were approved by the Animal Care Committee of Wenzhou Medical University.

A549 cells that had been transfected to stably express firefly luciferase (Xenogen, Alameda, CA) were infected with lentiviruses carrying MLL4 shRNA or scrambled control shRNA. These cells were injected into the lateral tail vein (5×10⁶ cells) of 6-week-old male nonobese diabetic/severe combined immunodeficiency (NOD SCID) mice. For bioluminescence imaging, mice were injected with 200 mg/g of D-luciferin in PBS abdominally. At 10 minutes after injection, mice were anesthetized and bioluminescence was imaged with a charge-coupled device camera (IVIS; Xenogen).

Results are shown as the mean±SD unless otherwise noted. Comparisons were performed using two-tailed paired t tests based on a bi-directional hypothesis for continuous variables.

To determine the expression of MLL4 in lung cancer, we analyzed the expression level of MLL4 in three published clinical datasets (GSE33356, GSE19188, and GSE18442) from the Gene Expression Omnibus (GEO) database (Fig. 1A). The results revealed that MLL4 expression was downregulated in lung cancer tissues compared with normal lung tissues. In addition, we found the mutation frequency of MLL4 was high in different lung cancer type databases (Fig. 1B). To confirm the result, we collected 16 pairs of NSCLC and adjacent normal tissues and found that the mRNA level of MLL4 was downregulated in the tumor tissues (Fig. 1C). Immunohistochemical analysis also revealed lower expression in the tumor tissues than the adjacent normal tissues (Fig. 1D). Then, we explored the protein expression level of MLL4 in 35 pairs of NSCLC and adjacent normal tissues with TNM clinical classification using immunohistochemistry staining (Fig. 1E). The results indicated a decreased expression level of MLL4 in the tumor tissues compared to the adjacent normal tissues, and its expression tended to be negatively correlated with disease stage progression (Fig. 1F and G).

To determine how MLL4 regulates non–small cell lung carcinogenesis, we investigated the transcriptomes in the control and MLL4 siRNA A549 cells using a high-throughput RNA deep sequencing approach (RNA-seq). Total RNA was extracted from A549 cells transfected with control siRNA or siRNA-targeting MLL4, and three independent samples and controls were assessed. Whole-transcriptome clustering analysis revealed that the expression levels of 1,042 genes were significantly changed upon MLL4 knockdown (Fig. 2A). A total of 515 genes showed up-regulated expression, whereas 527 genes showed downregulated expression in the A549 cells treated with MLL4 siRNA compared with samples treated with control siRNA. Gene set enrichment analyses showed significant enrichment in transforming growth factor β (TGF-β) signaling pathways after MLL4 knockdown (Fig. 2B). Furthermore, the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were conducted based on both genes with up-regulated and downregulated expression, revealing the vital 12 enriched pathways. These signaling pathways include the PI3K-AKT, tumor necrosis factor, advanced glycation end-products–receptor for advanced glycation end products, TGF-β, and interleukin-17 signaling pathways (Fig. 2C). To validate the RNA-seq analysis, we chose 19 representative genes (Fig. 2D) and validated their expression in A549 and SK-MES-1 cells using qPCR. The results indicated that the mRNA levels of ATF1, FOXP4, KDM2A, SOX2, SOX21, PABL6, SMAD9, BMP6 and THBS1 increased, while the RAB37, RASSF4, PACR3, PIK3IP1, SIRT4, TENT5B, LFNG, RBL1, BATF2 and CXXC5 levels decreased upon the knockdown of MLL4 (Fig. 2E and F). Western blot demonstrated that PIK3IP1 expression was downregulated following MLL4 knockdown, whereas the P-AKT, SOX2, and THBS1 levels were up-regulated and the expression of AKT was unchanged (Fig. 2G). To determine whether MLL4 could affect the global level of H3K4me1 in NSCLC cells, cells were transfected with siRNA specifically targeting MLL4. The results showed that knockdown of MLL4 led to a decreased expression of global H3K4me1 level in A549 and SK-MES-1 cells (Fig. 2H). Since MLL4 regulates gene expression mainly through binding and methylation of enhancer regions, we also performed ChIP assay using antibodies against MLL4 or control IgG. The results revealed that MLL4 does not occupy at the classical enhancers (SSR1 and SSR2) [20] and promoters (SOX2-promoter) region of SOX2 gene (S4A Fig.). In order to determine whether the regulation of SOX2 by MLL4 was dependent on PIK3IP1. Using western blotting, we found that the increased expression of SOX2 with MLL4 knockdown, which could be restored by combining with PIK3IP1 overexpression (Fig. 2I). In addition, we also detected the expression of SOX2 in the MLL4 knockdown A549 and SK-MES-1 cells by treating the PI3K inhibitor (ZSTK474) (Fig. 2J). The results showed that the expression of SOX2 was significantly decreased after treatment with PI3K inhibitor in both control siRNA and MLL4 siRNA cells. These results suggested that MLL4 regulates the expression of SOX2 in PIK3IP1-dependent manner and MLL4 may regulate the progression of NSCLC by regulating the PI3K/AKT/SOX2 axis.

To investigate the function of MLL4 in NSCLC, we evaluated cell proliferation in the NSCLC cells. We used CRISPRa to activate the expression of MLL4 and used the lentivirus-mediated stable knockdown efficiencies of MLL4 by two MLL4-specific shRNA (#1 and #2). The results of EdU staining demonstrated that the percentage of EdU-positive cells was significantly decreased in MLL4 overexpression A459 cells and increased in the MLL4-silenced A549 cells. Furthermore, in agreement with the functional link between MLL4 and PIK3IP1 described previously, the increase in EdU-positive cells resulting from MLL4 knockdown was partially reversed by overexpression of PIK3IP1 (Fig. 3A). In addition, consistent results were obtained in SK-MES-1 cells (Fig. 3B). Colony formation assays further showed that shRNA knockdown of MLL4 increased the colony formation ability, while could be restored by combining with PIK3IP1 overexpression in A549 cells and SK-MES-1 cells (Fig. 3C and D). We confirmed the efficiency of lentivirus-mediated shRNAs targeting MLL4 and CRISPRa-MLL4 gRNA using reverse transcription polymerase chain reaction (RT-PCR), and the expression level of PIK3IP1 overexpression was detected by western blot analysis (Fig. 3E). Moreover, EdU staining assays (Fig. 4A and B) and colony formation assays (Fig. 4C and D) also showed that knockdown of PIK3IP1 or overexpression of SOX2 mimics the phenotype of MLL4 knockdown in A549 cells and SK-MES-1 cells. The western blots shown in Fig. 4E further verified the efficacies of the shRNAs and plasmids used in these experiments. In addition, growth curve assays also showed that MLL4 knockdown significantly increased the growth of A549 cells and SK-MES-1 cells, which could be partially alleviated by overexpression of PIK3IP1 (Fig. 5A and B).

Next, we investigated the role of MLL4 in tumor migration and invasion. For this purpose, CRISPRa was used to activate the expression of MLL4 and MLL4 knockdown by MLL4-specific shRNA in A549 and SK-MES-1 cells. In the wound-healing assays, the amount of open distance remaining after 36 hours of migration was different compared to that of the control. The overexpression of MLL4 resulted in a significant delay in wound closure in A549 and SK-MES-1 cells, whereas MLL4 knockdown was associated with an increased migration rate, which could be partially rescued by overexpression of PIK3IP1 (Fig. 5C and D). Furthermore, the results from transwell invasion assays in two NSCLC cell lines (A549 and SK-MES-1) showed a significantly decreased invasiveness caused by MLL4-CRISPRa. Moreover, we also found that a significant increase in invasiveness was associated with MLL4 knockdown, which could be restored by combining with PIK3IP1 overexpression (Fig. 5E and F). RT-PCR and western blot shown in Fig. 5G further verified the efficacies of the indicated shRNAs, CRISPRa-MLL4 gRNA, and plasmids used in these experiments. Meanwhile, PIK3IP1 knockdown and SOX2 overexpression showed increased migration rate (Fig. 6A and B) and invasiveness (Fig. 6C and D) compared to the control groups, consistent with the phenotype of MLL4 knockdown. The western blots shown in Fig. 6E further verified the efficacies of the shRNAs and plasmids used in these experiments. These data are consistent with a role of MLL4 in suppressing carcinogenesis and support the observation that PIK3IP1 is a downstream target of MLL4.

Sphere formation assay is an in vitro method commonly used to identify CSCs and study their properties. Since SOX2, the target gene regulated by MLL4, has an important role in maintaining stem cell properties and functions, we evaluated the role of MLL4 in the sphere forming ability of A549 and SK-MES-1 cells. The results showed that knockdown of MLL4 increased the number and size of the spheres of A549 (Fig. 7A) and SK-MES-1 cells (Fig. 7B) compared to the control cells in low-attachment plates. Moreover, the increased size and number of spheres with MLL4-targeted shRNA was markedly blocked by co-transfected with PIK3IP1. The expression level of MLL4 and PIK3IP1 was further verified through RT-qPCR and Western blot (Fig. 7C). In addition, knockdown of SOX2 decreased the number and size of the spheres, and PIK3IP1 knockdown and SOX2 overexpression showed increased the number and size of the spheres in A549 (Fig. 8A) and SK-MES-1 cells (Fig. 8B). The western blots shown in Fig. 8C further verified the efficacies of the shRNAs and plasmids used in these experiments. Based on these results, low expression of MLL4 may be required for the maintenance of NSCLC stem cell properties, and it does so, at least in part, through regulating the PI3K/AKT/SOX2 axis.

Furthermore, we investigated the role of MLL4 in tumor development and progression in vivo by subcutaneously implanting A549 cells that had been engineered to stably express MLL4 shRNA or control scrambled shRNA in athymic BALB/c mice. Growth of the implanted tumors was monitored by measuring the tumor sizes every 3 days over a period of 4 weeks. The results showed that tumor growth was substantially increased upon MLL4 knockdown (Fig. 9A). The average tumor volume/weight of mice in the MLL4 interference group was 2.59/1.93 times that of the control group. To further explore the role of MLL4 in NSCLC progression. We investigated the role of MLL4 in NSCLC metastasis in vivo. A549 cells that had been engineered to stably express firefly luciferase (A549-Luc) were infected with lentiviruses carrying shRNA against MLL4 and then injected intravenously through the tail vein of 6-week-old male NOD SCID mice. Lung metastasis was quantified using bioluminescence imaging after 4 weeks. The results showed that MLL4 knockdown led to a dramatic increase in lung metastasis of the A549-Luc tumors. The metastases to the lungs were verified by bioluminescence imaging and histological staining (Fig. 9B). In order to further extend our observation results to a clinic-pathologically relevant context, we analyzed the expression of MLL4, PIK3IP1, and SOX2 with clinical behavior in lung cancer patients. Kaplan-Meier survival analysis (http://kmplot.com/analysis/) shows that the overexpression of MLL4/PIK3IP1 or low expression of SOX2 was associated with improved overall survival in the lung cancer patients (Fig. 9C). In addition, a reanalysis of the data sourced from published clinical datasets such as GSE108492, GSE115457, GSE8894, and GSE3141 indicated that the expression of MLL4 is significantly positively correlated with PIK3IP1 expression and negatively correlated with SOX2 expression, supporting our finding that PIK3IP1 was transcriptionally regulated by MLL4 (Fig. 9D). Taken together, these findings indicated that low expression of MLL4 promotes NSCLC cell tumorigenesis in vivo.